10.1007@978 981 15 8820 4

Algorithms for Intelligent Systems
Series Editors: Jagdish Chand Bansal · Kusum Deep · Atulya K. Nagar
Mohammad Shorif Uddin
Avdhesh Sharma
Kusum Lata Agarwal
Mukesh Saraswat Editors
Intelligent
Energy
Management
Technologies
ICAEM 2019
Series Editors
Jagdish Chand Bansal, Department of Mathematics, South Asian University,
New Delhi, Delhi, India
Kusum Deep, Department of Mathematics, Indian Institute of Technology Roorkee,
Roorkee, Uttarakhand, India
Atulya K. Nagar, School of Mathematics, Computer Science and Engineering,
Liverpool Hope University, Liverpool, UK
This book series publishes research on the analysis and development of algorithms
for intelligent systems with their applications to various real world problems. It
covers research related to autonomous agents, multi-agent systems, behavioral
modeling, reinforcement learning, game theory, mechanism design, machine
learning, meta-heuristic search, optimization, planning and scheduling, artificial
neural networks, evolutionary computation, swarm intelligence and other algo-
rithms for intelligent systems.
The book series includes recent advancements, modification and applications
of the artificial neural networks, evolutionary computation, swarm intelligence,
artificial immune systems, fuzzy system, autonomous and multi agent systems,
machine learning and other intelligent systems related areas. The material will be
beneficial for the graduate students, post-graduate students as well as the
researchers who want a broader view of advances in algorithms for intelligent
systems. The contents will also be useful to the researchers from other fields who
have no knowledge of the power of intelligent systems, e.g. the researchers in the
field of bioinformatics, biochemists, mechanical and chemical engineers,
economists, musicians and medical practitioners.
The series publishes monographs, edited volumes, advanced textbooks and
selected proceedings.
More information about this series at http://www.springer.com/series/16171

Mohammad Shorif Uddin Avdhesh Sharma
• •
Kusum Lata Agarwal Mukesh Saraswat

•
Editors
Intelligent Energy
Management Technologies
ICAEM 2019
123
Editors
Mohammad Shorif Uddin Avdhesh Sharma
Department of Computer Science Department of Electrical Engineering
and Engineering MBM Engineering College
Jahangirnagar University Jodhpur, India
Dhaka, Bangladesh
Mukesh Saraswat
Kusum Lata Agarwal Jaypee Institute of Information Technology
Department of Electrical Engineering Noida, India
Jodhpur Institute of Engineering
and Technology
Jodhpur, India
ISSN 2524-7565 ISSN 2524-7573 (electronic)

ISBN 978-981-15-8819-8 ISBN 978-981-15-8820-4 (eBook)
https://doi.org/10.1007/978-981-15-8820-4
© Springer Nature Singapore Pte Ltd. 2021
This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part
of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations,
recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission
or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar
methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this
publication does not imply, even in the absence of a specific statement, that such names are exempt from
the relevant protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this
book are believed to be true and accurate at the date of publication. Neither the publisher nor the
authors or the editors give a warranty, expressed or implied, with respect to the material contained
herein or for any errors or omissions that may have been made. The publisher remains neutral with regard
to jurisdictional claims in published maps and institutional affiliations.
This Springer imprint is published by the registered company Springer Nature Singapore Pte Ltd.
The registered company address is: 152 Beach Road, #21-01/04 Gateway East, Singapore 189721,
Singapore
Preface
Twenty-first century is the transition from fossil-fuel-based energy to renewable

energy, and after Montreal and Kyoto Protocol, the world is inclined toward renew-
able energy sources. Intelligent energy management technologies are now appearing
very cost-effective compared to the projected high cost of fossil fuels. In this context,
numerous intelligent control schemes are proposed for developing smart energy
management systems. With this background, the idea of the International Conference
on Advances in Energy Management System (ICAEM 2019) is conceived to develop
a platform for networking; disseminating; and exchanging challenges, ideas, con-
cepts, and results among the researchers from academia and industry.
This book contains the good-quality research papers as the proceedings of this
International Conference on Advances in Energy Management System (ICAEM
2019). ICAEM 2019 has been jointly organized by the Rajasthan Technical University
(RTU), Kota, and Jodhpur Institute of Engineering and Technology (JIET), Jodhpur,
Rajasthan, India. It was held on December 20–21, 2019 at JIET, Jodhpur, Rajasthan,
India. The conference focused on the application of artificial intelligence, soft com-
puting, optimization, machine learning, intelligent software, data science, data secu-
rity, and big data analytics on advanced energy management systems.
We have tried our best to enrich the quality of the ICAEM 2019 through a
stringent and careful peer-review process. ICAEM 2019 received 132 papers on
four conference tracks. Among these 65 papers were selected through the
peer-review process for presentation during the conference. However, the final
proceedings contain only 38 papers after careful editorial reviewing.
The book presents the intelligent computing research results in reliable power
systems, power quality, smart grids, energy management, conversion techniques,
energy economics, etc. We believe that this book serves as a reference material for
advanced research in the field of energy management.
Dhaka, Bangladesh Mohammad Shorif Uddin

Jodhpur, India Avdhesh Sharma
Jodhpur, India Kusum Lata Agarwal
Noida, India Mukesh Saraswat
v
Contents
1 Solar PV-Based Electric Vehicle Charging System

with Power Backup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Utkarsh Gupta, D. K. Yadav, and Dheeraj Panchauli
2 A Survey of Energy Theft Detection Approaches
in Smart Meters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
Divam Lehri and Arjun Choudhary
3 Renewable Energy Conversion: Sustainable Energy Development
and Efficiency Enhancement of Solar Panels: A Review . . . . . . . . . 25
Bhavik J. Pandya, Megha C. Karia, and Kamlesh B. Sangani
4 Evaluation of Thermal Degradation Behavior
of Cardboard Waste . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
Samit Kumar Singh and Sadanand A. Namjoshi
5 Wavelength Optimization in Gigabit Passive Optical Network
by Proposed Quad Play Architecture . . . . . . . . . . . . . . . . . . . . . . . 53
Md. Hayder Ali and Mohammad Hanif Ali
6 A Priority-Based Deficit Weighted Round Robin Queuing
for Dynamic Bandwidth Allocation Algorithm in Gigabit
Passive Optical Network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
7 Protection of Six-Phase Transmission Line Using Demeyer
Wavelet Transform . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
Gaurav Kapoor
8 Analysis of Voltage Sag and Swell Problems Using Fuzzy Logic
for Power Quality Progress in Reliable Power System . . . . . . . . . . 85
Ankit Tandon and Amit Singhal
vii
viii Contents
9 Vehicle Detection and Its Speed Measurement . . . . . . . . . . . . . . . . 95

Morium Akter, Jannatul Ferdous, Mahmuda Najnin Eva,
Sumaita Binte Shorif, Sk. Fahmida Islam,
and Mohammad Shorif Uddin
10 Fuel Cells as Naval Prime Movers: Feasibility, Advances
and Implications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
Akshat Mathur and Sushma Dave
11 Overview of Eco-Friendly Construction Materials . . . . . . . . . . . . . 111
Nidhi Sharma and Aashish Kumar Jha
12 Protection of Nine-Phase Transmission Line Using
Biorthogonal-2.2 Wavelet Transform . . . . . . . . . . . . . . . . . . . . . . . 119
Gaurav Kapoor
13 An Intelligent Hybrid Model for Forecasting of Heart
and Diabetes Diseases with SMO and ANN . . . . . . . . . . . . . . . . . . 133
Shalini, Pawan Kumar Saini, and Yatendra Mohan Sharma
14 Power Quality Assessment of Solar PV Standalone System
Using Various DC-DC Converters . . . . . . . . . . . . . . . . . . . . . . . . . 139
Surbhi Shringi, Santosh Kumar Sharma, and Utkarsh Gupta
15 Coordinated Control of UPFC-Based Damping Controller
with PID for Power System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
Amit Singhal and Ankit Tandon
16 Computational Neuroscience and Its Applications: A Review . . . . . 159
Aisha Jangid, Laxmi Chaudhary, and Komal Sharma
17 Optimization of Band Notch Characteristic in Ultra-Wideband
Microstrip Patch Antenna for Wireless Power Transfer . . . . . . . . . 171
Ashish Mathur, Geetika Mathur, and Abhijit kulshrestha
18 Experimental Investigation for Energy-Conscious Welding Based
on Artificial Neural Network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181
Sudeep Kumar Singh, Suvam Sourav Swain, Amit Kumar,
Prashanjeet Patra, Nitesh Kumar, and A. M. Mohanty
19 Detection of GSM Signal Using Energy Detection and Matched
Filter-Based Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191
Bablu Kumar Singh and Sanjay Bhandari
20 Simulation of Performance Characteristics of Different
PV Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199
Harish Kumar Khyani and Jayashri Vajpai
21 Stabilization Analysis of Clay Soil by Using Waste
Tile Particles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209
Saraswati Chand Dhariwal and Rajat Mangal
Contents ix
22 Renewable Energy as Biofuel from Mirabilis Jalapa Seed Oil . . . . 219

Akleshwar Mathur and Harish Kumar Khyani
23 A Comparative Study of Various Approaches to Lossy Image
Compression Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227
Nitesh Agarwal and Rajendra Purohit
24 Practical Design Considerations of DC/DC Converter Used
in MPPT for Solar PV Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . 251
Kusum Lata Agarwal and Dhanraj
25 Machine Learning-Based Solar Energy Forecasting: Implications
on Grid and Power Market . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 267
Chandershekhar Singh and Ajay Kushwaha
26 Performance Comparison of Minimum Shift Keying
and Gaussian Minimum Shift Keying in Additive White Gaussian
Noise Environment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277
Bablu Kumar Singh and Supriya Vyas
27 MATLAB-Based Comparative Analysis of Alternative
PV Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 285
Khamma Kanwar and Jayashri Vajpai
28 Nuclear Fusion: Energy of Future . . . . . . . . . . . . . . . . . . . . . . . . . 295
Jayesh Nehiwal, Harish Kumar Khyani, Shrawan Ram Patel,
and Chandershekhar Singh
29 Medical Image Processing Using Soft Computing Techniques
and Mathematical Morphology . . . . . . . . . . . . . . . . . . . . . . . . . . . . 303
Pratik Bhansali and Sandip Mehta
30 Analysis of Hotspot Development in Power Transformer
and Its Life Estimation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 319
Vinit Mehta and Jayashri Vajpai
31 Photovoltaic Module Cleaning Prediction Using Deep
Neural Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 335
Kapil Panwar, Aditya Jindal, and Kusum Lata Agarwal
32 Optimal Controller Design for DC–DC Buck Converter . . . . . . . . . 343
Shubham Sharma and Kusum Lata Agarwal
33 Comparative Analysis of Different Maximum Power Point
Techniques for Solar Photovoltaic Systems . . . . . . . . . . . . . . . . . . . 357
Shyam Lal Vishnoi and Kusumlata Agarwal
34 MATLAB/Simulink-Based Tracking of Maximum Power Point
in a Generalized Photovoltaic Module by Using DC-DC
Boost Converter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 381
Yogesh Joshi and Vinit Mehta
x Contents
35 Design and Simulation of MPPT-Operated DC-DC Flyback

Converter Used for Solar Photovoltaic System . . . . . . . . . . . . . . . . 397
Ranjana Choudhary and Shrawan Ram Patel
36 To Improve Power Transfer Capacity Using TCSC FACTS
Controller . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 409
Kishore Singh Gehlot and Shrawan Ram
37 Stockwell Transform and Hilbert Transform Based Hybrid
Algorithm for Recognition of Power Quality Disturbances . . . . . . . 425
Ramesh Aseri and Ashwani Kumar Joshi
38 Detection and Classification of Transmission Line Faults Using
Combined Features of Stockwell Transform, Hilbert Transform,
and Wigner Distribution Function . . . . . . . . . . . . . . . . . . . . . . . . . 449
Tanmay Bhati and Harish Kumar Khyani
Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 469

About the Editors
Dr. Mohammad Shorif Uddin is a Professor of Computer Science and

Engineering at Jahangirnagar University, Bangladesh. He completed his Bachelor
of Science in Electrical and Electronic Engineering at BUET in 1991, Master of
Technology Education at Shiga University, Japan, in 1999, Doctor of Engineering
in Information Science at Kyoto Institute of Technology, Japan, in 2002, and an
M.B.A. at Jahangirnagar University in 2013. He undertook postdoctoral researches
at Bioinformatics Institute, Singapore, Toyota Technological Institute, Japan, Kyoto
Institute of Technology, Japan, Chiba University, Japan, Bonn University,
Germany, and Institute of Automation, Chinese Academy of Sciences, China. His
research is motivated by applications in the fields of artificial intelligence, imaging
informatics, and computer vision. He is the Editor-in-Chief of ULAB Journal of
Science and Engineering and an Associate Editor of IEEE Access and has served as
General Chair or Co-Chair of various conferences, including the IJCCI 2018 and
2019, EICT 2017, and IWCI 2016. He holds two patents for his scientific inven-
tions, is a senior member of several academic and professional associations, and has
published extensively in reputed international journals and conference proceedings.
Dr. Avdhesh Sharma is a Professor of Electrical Engineering at M.B.M. Engineering

College, India. He completed his Ph.D. in Electrical Engineering at IIT, Delhi, M.Tech.
in Computer Science & Data Processing at IIT, Kharagpur, M.Sc. at AMU, and B.Sc. in
Electrical Engineering at D.E.I. Engineering College, India. He has over 33+ years of
experience in teaching and research. He has published around 75 papers in international
journal and conference proceedings. He guided a good number of PG and Ph.D.
students. He is a Fellow of the Institute of Engineers, India.
Dr. Kusum Lata Agarwal is a Professor and Head of Electrical Engineering at

JIET, Jodhpur, India. She completed her Ph.D., M.E., and B.E. degrees at M.B.M.
Engineering College, India. She has over 18 years of experience in teaching and
research. She published around 30 papers in journal and conference proceedings.
xi
xii About the Editors
She supervised a good number of PG theses. She played a key role in organizing 4
national and international conferences and also achieved notable research grants.
She is in the reviewer panel of many journals and conferences.
Dr. Mukesh Saraswat is an Associate Professor of Computer Science & Engineering

at Jaypee Institute of Information Technology, India. He completed his B.Tech. in CSE
at MJP Rohilkhand University, M.Tech. in CSE at UPTU, and Ph.D. in CSE at IIITM,
Gwalior, India. He has more than 15 years of teaching and research experience. He has
guided more than 72 UG & PG theses and presently guiding 4 Ph.D. students. He is a
member of several academic and professional bodies and has published more than 35
papers in journals and conference proceedings in the area of imaging, pattern recog-
nition, and soft computing.
Chapter 1
Solar PV-Based Electric Vehicle
Charging System with Power Backup
Utkarsh Gupta, D. K. Yadav, and Dheeraj Panchauli
1 Introduction
With the rapid increase in technologies and popularity of EV, a need develops for
the improved charging infrastructure for their successful propulsion [1].
The charging could basically be powered by electricity generated from either
conventional or non-conventional source of energy. But to make the concept of EVs
completely environment-friendly a charging station powered by a renewable source
of energy is considered to be the best [2].
Out of many types of renewable sources present on earth like tidal, geothermal,
solar, wind energy, etc., the use of solar PV array for the EV charging station
is preferred the most due to its easy availability, ease of installation, and less
maintenance due to the absence of moving parts [3, 4].
The solar-powered charging system with power backup provides various advan-
tages to the infrastructure where it is installed as it provides free fuel for the EVs
throughout its lifetime, 20–25 years approximately, after a single investment and also
eliminates the need of the power backup sources like diesel generator, inverters, etc.
as it can supply the load by the EV batteries under emergency conditions [5].
U. Gupta (B) · D. K. Yadav · D. Panchauli

Rajasthan Technical University, Kota 324010, India
e-mail: Utkarshgupta3095@gmail.com
D. K. Yadav
e-mail: dkyadav@rtu.ac.in
D. Panchauli
e-mail: dheerajpanchauli777@gmail.com
© Springer Nature Singapore Pte Ltd. 2021 1

M. Shorif Uddin et al. (eds.), Intelligent Energy Management Technologies,
Algorithms for Intelligent Systems,
https://doi.org/10.1007/978-981-15-8820-4_1
2 U. Gupta et al.
2 System Description
See Fig. 1.
2.1 PV Array
A solar PV array of 1 Soltech 1 STH-215-P is used with 10 series modules and 40

parallel strings. The PV and IV characteristics of the PV array are as follows (Figs. 2).
Characteristics of PV array
See Fig. 3.
2.2 Boost Converter
A boost converter is a DC–DC converter that steps up voltage (while stepping down
the current) from its input to output. It is a category of SMPS (switch mode power
supply) containing at least two semiconductors (a diode and a transistor) and at least
one energy storage element, a capacitor, an inductor, or two in combination (Fig. 4).
2.3 Single-Phase Full Bridge Inverter for R-L Load Inverter
A single-phase square wave type voltage source inverter produces square-shaped

output voltage for a single-phase load. Such inverters have very simple control logic
and the power switches need to operate at much lower frequencies compared to
switches in some other types of inverters (Fig. 5).
Gate
pulse
25
AC
DC/DC Load
supply
800
PV-Array
EV
EV
Fig.1 Block diagram of the system

1 Solar PV-Based Electric Vehicle Charging System with Power Backup 3
Array type: 1Soltech 1STH-215-P; 10 series modules; 40 parallel strings

400
300
Current (A)
200
100
o
25 C
0
0 50 100 150 200 250 300 350 400
Voltage (V)
4
10
10
8 o
25 C
Power (W)
0
0 50 100 150 200 250 300 350 400
Voltage (V)
Fig. 2 The PV and IV characteristics of the PV array are as follows
500
MAGNITUDE
-500
0 0.5 1 1.5
VOLTAGE
400
MAGNITUDE
200
0
0 0.5 1 1.5
CURRENT
1000
MAGNITUDE
500
0
0 0.5 1 1.5
IRRADIANCE
26
MAGNITUDE
25
24
0 0.5 1 1.5
TEMPERATURE
Fig. 3 The output waveform of the PV array used in this system

4 U. Gupta et al.
Diode
DC C
Switch VOUT
Vin
Fig. 4 Basic diagram of a boost converter
Fig. 5 Basic diagram of a single-phase inverter
2.4 Single-Phase AC Supply
A supply of peak voltage of 180 V and frequency of 50 Hz is used to supply the load
under normal conditions.
2.5 Load
A resistive load of 5000 ohms is taken in this system.
3 Working of the System
The working of this system is explained basically in three modes.

Fig. 6 Simulation model of the system in Matlab
Fig. 7 Output waveforms across the load
MODE-I In this mode, both the grid and the SPV are ON where the grid supplies
the load and the SPV charges the batteries of electric vehicles.
MODE-II In this mode, the grid is absent and the SPV maintains the supply
across the load through a single-phase inverter and also charges the batteries of
the EVs.
MODE-III In this mode, both the PV and the grid are absent and the batteries of
the EVs supply the load and their discharging could be seen (Figs. 6, 7 and 8).
4 Simulation Model and Results
In Mode 1, i.e., from a time period of 0–0.5, the output waveforms across the load
show sinusoidal waveforms as in this period of time the load is supplied by the grid
6 U. Gupta et al.
Fig.8 Output waveforms across the battery
and the output waveforms across the battery show an increasing value of SOC as it
is getting charged by the solar PV array at this time.
In Mode 2, i.e., from time period of 0.5–1, the output waveforms across the load
can be seen as a square wave as it is supplied by the solar PV inverter via a single-
phase inverter in the absence of grid and the output waveforms across the battery
show the increasing SOC as it is getting charged by the solar PV array.
In Mode 3, i.e., from time period of 1–1.5, the output waveforms across the load
show the square wave of 144 V approximately which is supplied by the battery via
a single-phase inverter due to the absence of both the sources (solar and grid) and
discharging of the batteries can be seen in the output waveforms across the battery.
5 Conclusion
The charging station developed in this system proves to be a very economical system
as with eliminating the requirements of expensive power backup sources. It also
provides free fuel to the EVs throughout its life. The system could be further expanded
and could be made more useful by applying some control techniques.
References
1. Suman N, Yadav DK, Jangid T (2018) Modeling and analysis of photovoltaic system with
improved inverter technique. In: 2018 9th international conference on computing, communica-
tion and networking technologies (ICCCNT). https://doi.org/10.1109/icccnt.2018.8494168
2. Koduri N, Kumar S, Uday kumar RY (2014) On-board vehicle-to-Grid (V2G) integrator for
power transaction in the smart grid environment. In: 2014 IEEE international conference on
computational intelligence and computing research. https://doi.org/10.1109/iccic.2014.7238404
3. Chen J, Zhang Y, Su W (2015) An anonymous authentication scheme for plug-in electric vehicles
joining to charging/discharging station in vehicle-to-Grid (V2G) networks. China Commun
12(3):9–19. https://doi.org/10.1109/cc.2015.7084359
4. Yang P, PengT, Wang H, Han H, Yang J, Wang H (2017) A single-phase current-source bidirec-
tional converter for V2G application. In: 2017 IEEE 3rd international future energy electronics
conference and ECCE Asia (IFEEC 2017 - ECCE Asia). https://doi.org/10.1109/ifeec.2017.799
2125
5. Li Y, Yu G, Liu J, Deng F (2017) Design of V2G auxiliary service system based on 5G technology.
In: 2017 IEEE conference on energy internet and energy system integration (EI2). https://doi.
org/10.1109/ei2.2017.8245513
6. Han H, Lv Z, Huang D, Li Q (2017) Research on charge and discharge power tracking control for
V2G system. In: 2017 IEEE 2nd information technology, networking, electronic and automation
control conference (ITNEC). https://doi.org/10.1109/itnec.2017.8284928
Chapter 2
A Survey of Energy Theft Detection
Approaches in Smart Meters
Divam Lehri and Arjun Choudhary
1 Introduction
Since last decade numerous efforts have been made by the governments and distribu-
tion companies to counter electricity theft but it still remains a challenge. The entire
loss suffered by the power sector is known as Transmission and Distribution Losses
(TD Losses) which comprises an aggregate of Technical Losses (TL) and Non-
Technical Losses (NTL). TD losses represent the difference between the electricity
generated and the electricity consumed. Technical losses are those losses which are
internal to the system such as energy dissipation by the electrical equipments used in
distribution lines, transformers, transmission lines, and iron losses in transformers.
On the other hand, NTL constitutes losses arising due to defective meters, errors
in billing, flaws in supply, unmetered connections, and malicious activities by the
consumer such as tampering of meter. Table 1 provides an overview of different types
of electricity losses caused by different components of power sector.
The easiest way to determine the amount of non-technical losses (NTL) is by
merely calculating the technical losses (TL) in the system and subtracting it from
total losses (TD).
We can evaluate it as follows:
NTL = Total Energy Losses(TD) − TL (1)
Total Energy Losses = Energy Supplied − Bills paid (2)
D. Lehri (B) · A. Choudhary

Sardar Patel University of Police, Security and Criminal Justice, Jodhpur, India
e-mail: lehridivam@gmail.com
A. Choudhary
e-mail: a.choudhary@policeuniversity.ac.in

https://doi.org/10.1007/978-981-15-8820-4_2
10 D. Lehri and A. Choudhary
Table 1 Classification of methods of electricity theft

Elements Methods of theft
Meters Bypassing the meter
Deliberately damaging the meter seals or removing of the meter
Wires/Cables Illegal tapping to bare wires or underground cables
Transformers Illegal tapping of transformer terminals and junction boxes of overhead
lines
Billing irregularities Errors made by meter readers
Unpaid bills Unpaid bills by individuals or institutions
Some of the losses such as TL are unavoidable. The energy theft in India is majorly
due to unmetered usage of electricity. The concept of Transmission and Distribution
(TD) losses has been extended further to Aggregate Technical and Commercial losses
(AT&C).
AT & C Losses = {1 − (BE × CE)} × 100 (3)
T & D Losses = {1 − (BE)} × 100 (4)
where
Billing efficiency (BE) = Total unit Billed/Total unit Inputs (5)
Collection efficiency (CE) = Revenue collected/Amount Billed (6)
TD loss is the difference in input energy and energy billed. There is no account for
the losses arising due to low collection. AT&C loss is the difference in input energy
and energy for which revenue has been collected. Simply stated AT& C Loss can be
aggregated as
AT & C Losses = TL + CL (7)
Statistics on electricity losses in India shows that around 10–12% of AT&C losses
amount to technical reasons, while remaining 18–20% comprises commercial reasons
[1] known as commercial losses (CL). According to U.S. Energy Information Admin-
istration (EIA) [2] in the countries with low rate of theft and optimal technical
efficiency TD losses generally span between 6 and 8%. Figure 1 shows graphical
representation of AT &C loss percentage of different Indian states.
In such scenarios adoption of smart meters by the government of India could prove
as a game changer to curb electricity theft. There is also a grave need to develop
a common framework in the country where governments, manufacturers, research
institutions, Distribution system operators (DSOs) and academia work with mutual
2 A Survey of Energy Theft Detection Approaches in Smart Meters 11
Fig. 1 A visual representation of state-wise AT&C losses (%) for the period Apr’14–Mar’15,
According to catalog available on Open Government Data Platform (OGD), India [3]
cooperation to ensure resiliency, privacy and security of smart grid. A paradigm of

one such framework is SEGRID Project [4] for European Digital Grid. While smart
meters may not be completely theft resistant but they are capable of minimizing
the number of theft cases due to their immunity against traditional electricity theft
methods as well as the real-time monitoring of data between the utility companies
and the consumers.
Due to the complex architecture and large attack surface of Advanced Metering
Infrastructure (AMI), Smart Meters are vulnerable to tampering thereby requiring
effective theft prevention and detection techniques. In this paper, we present a survey
of available energy theft detection techniques.
2 Meter Tampering Methodologies
There are various mechanisms through which an adversary can tamper Smart Meters.
Methods of meter tampering can be divided into four classes:
• Current related tampering methods.
• Voltage related tampering methods.
• Mechanical tampering methods.
• Tampering by hacking and altering the memory.
Fig. 2 Actual connection
Fig. 3 Swapping of phase and neutral lines
A summary of mechanisms that are generally used to tamper smart meters is

presented in this section.
2.1 Swapping of Phase and Neutral Lines
In this method of tampering the adversary interchanges the phase and neutral lines.
This swapping of phase and neutral lines reverses the energy flow thereby effecting
the billing calculation (Figs. 2 and 3).
2.2 Double Feeding
Double feeding as the name suggests is a meter bypassing technique where an addi-
tional feeder is connected to the meter in such a manner that meter gets bypassed and
the energy consumption is not accounted for. Under such scenario the consumption
for the load affixed to the supplementary feeder won’t be recorded by the meter even
if the connection is legitimate. This type of tampering is generally done to connect
any heavy electric appliance so that it’s consumption remain unnoticed (Fig. 4).
Fig. 4 Double feeding
Fig. 5 Actual condition
2.3 Neutral Missing
In this method of meter tampering, the neutral line is completely cut-off from the
meter thereby resulting in zero input voltage. Hence the power computed by the
meter is zero (Since P = V * I and for given condition V = 0, therefore P = 0) This
tampering method is also referred to as single wire operation [5] (Figs. 5 and 6).
2.4 Neutral Disturbance
In this method of tampering some noise (High-Frequency voltage signals) is added

to the neutral line of the meter by connecting it through diode/variable resis-
tance/capacitor. The neutral of the meter gets deviated from its original point and
becomes unbalanced leading to less voltage recording by the meter and therefore
less energy consumption is recorded by the meter (Fig. 7).
Fig. 6 Neutral missing
Fig. 7 Neutral disturbance
2.5 Current Reversal by Connecting Input and Output

in Reverse
In this tampering event, the adversary connect the phase and neutral wires to the
wrong inputs. This causes the current to change direction from it’s original path in
which it was intended to flow. The intention of this kind of tampering is to dupe the
billing computation by reversing the route of current flow [6] (Fig. 8).
2.6 Partial Earth Fault Condition
It is a tampering method in which the load is connected to the earth due to which the
return current going back to the meter is reduced. This generates a difference in the
current stream flowing through the neutral wire and phase wire leading to current in
Fig. 8 Current reversal
Fig. 9 Partial fault connection
neutral wire become less than the current in the phase wire. Under normal conditions,
the current in the phase wire and the neutral wire is equal (Fig. 9).
2.7 Tampering Using High Frequency/Voltage
In this type of meter tampering a remote-controlled device is placed in close proximity

to the meter. The device is capable of generating high-electrostatic discharge. The
discharge so generated causes a spark in the meter thereby thwarting the meter from
recording the electricity consumption. Such method of theft is a concern as it does
not leave any trace or evidence.
Fig. 10 Classification tree for energy theft methods
2.8 Mechanical Tampering
Mechanical tampering includes methods where the meter is physically damaged in

order to record less or no energy usage. In such type of tampering the electrical
characteristics of the components of the meter are altered. Some of the conducts that
amount to such type of tampering are (Fig. 10)
• Opening of the meter covers by fracturing the meter seals.
• Subjecting meter body to chemicals.
• Subjecting meter to external magnetic field.
• Burning the meter.
• Using jammer devices.
3 Countermeasure Approaches Against Energy Theft
Along with the adoption of new technologies such as smart grid, a new era of attacks
are expected to emerge. The government and the utilities are now becoming aware
of these scenarios and are taking steps toward mitigating next generation of threats.
Rapid developments in the AMI have captured the attention of research organizations
and scholars from academia all around the sphere and a range of approaches have
been proposed to curb the menace of electricity theft. In this section, we will provide
a survey of the available approaches for energy theft detection.
3.1 Game Theory Based Detection Technique
In this technique, the stealing of electricity is represented as a play-off sandwiched

amidst the adversaries involved in electricity thief and the distribution utility. It is
a model projected on the concept of game theory where the main objective of the
adversary is to whip a predetermined amount of electricity and at the same time
minimizing the possibility of being identified, whereas the electricity utility desires
to augment the chance of detection of adversary and the level of operative price it
will sustain in administering this anomaly recognition operation [7]. However, it still
remains a challenge to construct a potential game plan and all players that include
regulators, thieves, and distributors.
Moreover, game theory is based on assumption that the number of players partic-
ipating in the game is finite. In country like India which is one of the largest in terms
of population, equipping smart meter in every household simply means a drastic
increase in number of players which makes game theory difficult to implement.
3.2 Supervised Learning Approach
In this approach load profiles for each customer is developed based upon the historical
data which is used as a classifier dataset. A pre-selection is made on the subset of smart
meters which are straightforwardly confirmed by the technicians within a specified
region and time. This process is carried out by the utility company and is referred to
as campaign [8]. Information such as consumption, profile, and external information
along with other parameters is used to design the profile. In general, a classification
problem of fraud identification is formulated which employs supervised learning
approach over the historical dataset of fraud cases that occurred in the past [8]. The
main criterion for evaluation is the (Odds Ratio) OR. OR may be computed between
the falsified clients against all the clients not incorporated in any campaign, called
as ORPG or between falsified clients and the non-falsified clients known as ORPN.
The ratios obtained from the campaign are mentioned in Table 1 in [8] and are based
on some of the characteristics obtained from the campaign. Based on probability
a fraud score is computed for each customer according to which the customer can
be classified as Fraudulent, Non-fraudulent, and Absent. However, this methodology
has performance challenges in scenarios where rate of campaigns is excessively high
or the size of campaigns is on a large scale.
3.3 Linear Error Correction Block Codes
Linear error-correcting block codes have a linear dependency between the bits of
input message and the parity bits. In other words, the resultant of sum of any two
codewords is also a codeword. At the receiving end, these bits are utilized to detect
and correct errors in the transmission. A computation of the total amount of power in
distinct combinations of the cables is computed repetitively and then these readings
are utilized to detect and correct errors in the meter readings [6]. In this approach,
the concept of syndrome decoding is applied where a generator matrix (G) is used
by sender to generate the codeword and decoding matrix, also called parity check
matrix (H) is used by the receiver to detect and correct the errors. If G.H = 0, then
the received codeword is correct. In case G.H = 0, we can determine the error using
the position of non zero bit and correct it. Additional meters, called check meters are
used to detect and correct single-bit errors in meter readings. It is assumed that there
are M check meters, which are capable of computing the sum of energies of desired
cable combinations [6]. However, this Linear block code detection mechanism is
prone to magnetic interference and can only detect that there is an error but could
not identify the actual meter on which the error exists.
3.4 Dynamic Programming Algorithm Based on Probabilistic

Detection
A novel algorithm based on dynamic programming which utilizes the tree struc-
ture of the distribution network has been proposed. It makes use of Feeder Remote
Terminal Unit (FRTU), which is capable of measuring analog and digital signals
and transmitting energy usage data back to the control unit wirelessly. The power
consumption in the downstream network commencing from FRTU can be tracked
down by means of the information gathered from FRTU. If the power consump-
tion varies notably compared to the aggregate of the readings of smart meters in
the downstream network than it is concluded that at least one of the meters has
been compromised [9]. The algorithm aims to install FRTUs in minimal quantity in
power distribution networks due to cybersecurity concerns. Moreover the algorithm
is optimized to increase efficiency by utilizing solution pruning techniques. The main
parameters that the algorithm uses to determine theft are the Attacking Probability
and the Anomaly Score which are defined in [9]. Based on the anomaly score it is
decided whether a meter is anomalous or not. The proposed algorithm works under
the supposition that the adversary can only attack the smart meter equipped in her/his
own apartment. It may not provide a fruitful solution in situations where adversary
uses advanced techniques such as remotely attacking meter in Neighborhood Area
Network (NAN). Another probabilistic approach has been proposed in [10] which
provides an estimate of Technical and Non-Technical Losses in a segregated manner.
3.5 Temperature-Dependent Predictive Model
“Temperature-Dependent Technical Loss Model (TDTLM)” is the advancement of

the “Constant Resistance Technical Loss Model (CRTLM)” [11] by making the
resistance temperature dependent. To estimate NTL, TDTLM utilizes the property
that there is a linear dependency between the resistance of material and its temperature
in. The power consumption values along with other instantaneous measurements are
aggregated and sent back to the utility repeatedly after a fixed interval of 30 min for
calculation of NTL. Based on the threshold value of NTL cases are classified as theft
and non-theft. To train the predictive model data from the first two days (no theft) is
utilized. Whenever NTL estimate exceeds the threshold value, it is suspected that a
power theft has occurred in the user group.
3.6 Current Bypass Anti-Tampering Algorithm
A single-chip solution has been developed where an anti-tampering algorithm has

been implemented on an “ARM Coretext-M3 (STM32L152VB)” microcontroller. It
is a low power microcontroller operating at 32 MHz using “ADE7953 (Single Phase
Smart Meter)” and “ADE7878 (Three Phase Smart Meter)” Analog Devices [12].
The unbalance current difference ( Irr) is calculated by extracting data values from
IRMSA (Ia ) and IRMSB (Ib ) registers of ADE7953, where current in phase line is
denoted as Ia and current in neutral line is denoted as Ib . The unbalance current
difference ( Irr) is expressed in Eq. (1) of [12] as
|Irr| = Ia − Ib ÷ (Ia + Ib ) (8)
Verification of current bypass tampering event by the smart meter is done by

comparing the calculated Irr value in (4) against the pre-defined threshold values
which are 2.5% in case of three-phase meter and 1% in case of single-phase meter
[12]. In case of any uneven event, an interrupt is sent to the MCU. The MCU verifies
the tampering event by examining the status bits in the registers of ADE7953 and
ADE7878.
3.7 Microprocessor-Based Theft Control System
The theft control system based on “ARM-Cortex M3 processor” has been imple-
mented to prevent the energy meter from tampering attacks such as disconnections
of phase/neutral lines, entire meter bypassing, and meter tampering. This approach
utilizes the current difference in phase line and neutral line to detect tampering event.
Two current transformers, one in each phase line and neutral line are inducted. In case
of any disconnection of either of the lines from the meter, it would result in a signifi-
cant drop in current measured by the each current transformer [13]. This difference is
computed by the microcontroller by measuring current through ADC. In case of any
irregular difference an SMS is sent to the electricity utility by the microcontroller.
This functionality can be integrated into existing meters in addition to manufacturing
it in new meters. The module uses GSM network for communication which is already
well established in India.
3.8 AMIDS Framework
Advanced Metering Infrastructure Intrusion Detection System (AMIDS) is a frame-

work developed using an amalgamation of a variety of approaches for detection
and reporting of energy theft in smart meters. An attack graph-based data fusion
algorithm is used by AMIDS to merge artefacts of on-going attacks from numerous
sources [14]. The attack graph so composed is a directed graph based on state which
consists of different stages from initial to final. To achieve information fusion online,
the attack graph is considered as a Hidden Markov Model (HMM). AMIDS makes
use of both a supervised learning methodology that can compute individual applica-
tion usage and an unsupervised methodology that learns by clustering load events.
AMIDS takes into account numerous information sources to collect adequate amount
of artefacts regarding an on-going attack prior to identifying an activity as a malicious
energy theft.
3.9 Model Based on Harmonic Generators
A model has been proposed in [15] which uses harmonic sensors to determine the
uncertainty in smart meter readings. The proposed model consists of harmonic sensor,
ICS, energy meter, circuit breaker, and communication system. The External Control
Station (ECS) situated at the utility company receives instant values from end user
side. The non-technical loss is calculated by ECS and in case of loss being more
than 5% it would break the supply to the meter by indicating the control system to
disconnect the customer. The core of the model is the harmonic sensor which compute
the uncertainty in meter reading based on Total Harmonic Distortion (THD). In [16]
this approach based on the harmonic generators is extended by placing two smart
meters along with harmonic sensors and generators at either ends. One meter is
placed at the consumer end and other one at the utility end which makes it possible
to keep track of the generated power as well as the consumed power.
The difference in generated and consumed power is calculated to determine the
theft.
3.10 MIDAS Framework
A novel framework MIDAS [17] is developed which is the integration of several tech-
niques such as statistical analysis, data mining, and neural network. This framework
is different from other approaches as it classifies not only suspected users but also
classifies users without technical losses. Data mining is performed for fault and theft
sensing and to analyze load profiles of individual customers. The neural network is
trained with multiple methods. Different neural network topologies are developed
and at end of training the Root Mean Square (RMS) value for each model is computed.
The model with minimum RMS value is presented as final neural network.
3.11 Measuring Voltage Drop Between Smart Meters
This approach utilizes the magnitude of the voltage drop between two smart meters
to detect and decrease illegal consumption. The concept is to grab the voltage and
power data from the meter. It functions on pre-condition that there should be more
than two consumers involved in the powering of transformer, because detection
of unauthorized spending is computed by comparing the drop of voltage of each
measuring point to the transformer. In case if a drop of voltage occurs than it is
deduced that the consumer is having unauthorized connection to the meter [18].
3.12 Energy Lens
Energy Lens system intelligently integrates electricity meter data with sensors on
smartphones. It is expected that users using energy lens possess an android phone
with the capability to sample microphone audio and WiFi signal strength. During
initial phase users are required to turn on electricity appliances which they want to get
recognized by the energy lens. Users wait for some time for its power consumption
to reach a steady state and then turn it off. Based on this data acquired, it is identified
that when, where, and by whom the activity is performed upon the execution of
algorithm on the server [19]. However, energy lens faces several challenges such
as acquisition of ground truth statistics for building up the precision, the effect of
phone’s direction and privacy of the customer.
3.13 FNFD (“Fast NTL Fraud Detrection and Verification”)
FNFD is a mathematical method constructed on the notion of Recursive Least Square

(RLS) to represent adversary behavior. Using FNFD the NTL fraud, in Smart Grid
is detected in real time. FNFD is capable of verifying a fraud even with one single
measurement, given that the historical data supplied to it is accurate [20]. FNFD
employs linear functions to simulate the behavior of adversary. The main advantage
of FNFD as compared to other schemes is that it requires much less data and supports
NTL fraud verification, a unique feature that is not available in other schemes and it
is much faster than the other similar frameworks.
4 Proposed Work
All the existing works available on energy theft detection in smart meters are dealing
only with types of thefts where by some means either phase or neutral wires were
swapped or removed which led to significant change in the voltage or current values
or due to billing irregularities. Our work extends the existing approaches to a new
threat scenario where theft detection in smart meter occurs due to tampering of the
hardware chip of the meter itself. Our approach is mainly concerned with the chip-
level tampering of smart meters. An adversary could Reverse engineer the meter and
obtain the low-level assembly instructions of the meter. Further disassembling of the
smart meter could be done, thereby attempting to read the firmware directly from
the chip. Obtaining of the low-level assembly instructions would reveal the hard-
coded cryptographic keys among other sensitive information that can be used in later
attacks. Moreover, by exploiting the low-level assembly code the adversary could
alter the consumption readings. The common methods of exploiting the hardware
chip include:
• Logical Analyzer.
• Circuit Bending.
• JTAG Method.
• Hacking Over UART.
The logical analyzer is an instrument which sniffs the signals when placed on
different test points on the circuit board thereby revealing potential information that
could be interpreted into something useful, adding or removing circuit components
such that the functionality of the circuit is affected, also known as Circuit Bending
and using Joint Test Action Group (JTAG) method to read full memory hex dump.
We will begin with exploring the embedded hardware of the smart meter, exam-
ining individual components present on the circuit board. To get a better under-
standing of the working of each component we will probe the datasheets associated
with each component. Extending our approach further we will examine the inter-
connections between different components using multi-meter. This will provide us
insight of how data and signal transmission is taking place on the device. Now we will
hunt for debug ports present on the device. JTAG and UART are the most common
debug ports and we can easily identify them by monitoring the voltage levels using
a multi-meter or with the help of oscilloscope. Once debug ports are identified we
will start interacting with the device by making connections between the debug ports
and any USB bridge. USB Bridge will provide us with the capability to interact
with the device through console and finally we will begin the process of extracting
data\firmware from the device. We will modify the data dump that we acquired and
rewrite it to the device such that we can manipulate the device.
In continuation to this paper, we will be showcasing this kind of meter tampering
using these mentioned method along with the experimental results. We will also
propose mitigation measures for such type of chip-level energy theft approach such
as assembly code obfuscation.
5 Conclusion
Curbing the energy theft menace is a huge concern for the governments and utilities.
The scope of tampering comprises straightforward approaches like controlling live
or neutral wires to more grave means like retrieving device firmware. Appliances like
smart meters are part of critical infrastructure and any compromise to it would be
causing chaos in the power sector and huge loss of revenue to the government. Most of
the critical infrastructure devices are procured from global sources and may come pre-
installed with hardware backdoors. Adversary can also intrude through the weakest
point in the supply chain and compromise the device by installing hardware backdoor.
This shows that attackers are now moving down the stack from application layer
attacks to embedded hardware of the devices. The tools required to carry out physical
attacks are also proliferating and becoming inexpensive. Such scenarios call for
importance to hardware-level security which is not usually considered as important
as application-level security. Organizations need to reshape their security approach
from the viewpoint of attackers and conduct red team assessments to enhance the
security of the assets. In recent years, the advancement of smart grid and adoption of
smart meters has called for proposals from industry, universities, and governments
to tackle the vulnerabilities existing in the AMI. In this paper, we have classified
various ways of energy theft and detection techniques along with their challenges.
However, it still remains a fresh topic and has a lot of room to be worked upon in the
future.
References
1. Sharma T, Pandey KK, Punia DK, Rao J (2016) Of pilferers and poachers: Com-bating
electricity theft in India. Energy Res Soc Sci 11:40–52
2. U.S. Energy Information Administration—Eia—Independent Statistics and Analysis. https://
www.eia.gov/todayinenergy/detail.php?id=23452
3. Aggregate Technical & Commercial (AT&C) Losses in power sector. saumitra.cea@nic.in.
https://data.gov.in/catalog/aggregate-technical-commercial-atc-losses-power-sector.
Accessed Dec 17
4. Securing the smart grid of tomorrow. https://segrid.eu/
5. Warudkar D, Chandel P, Salwe BA (2014) Anti-tamper features in energy meters. Int J Electr,
Electron Data Commun 2(5), May-2014. ISSN 2320-2084
6. Mesbah W (2016). Detection and correction of tampering attempts of smart electricity meters.
In: PES innovative smart grid technologies conference Europe (ISGT-Europe), October, pp
1–6. IEEE
7. Cárdenas AA, Amin S, Schwartz G, Dong R, Sastry S (2012) A game theory model for elec-
tricity theft detection and privacy-aware control in AMI systems. In: Communication, control,
and computing (Allerton), 2012 50th annual Allerton conference on, October, pp 1830–1837.
IEEE
8. Coma-Puig B, Carmona J, Gavalda R, Alcoverro S, Martin V (2016) Fraud detection in energy
consumption: a supervised approach. In: Data science and advanced analytics (DSAA), 2016
IEEE international conference on, October, pp 120–129. IEEE
9. Zhou Y, Chen X, Zomaya AY, Wang L, Hu S (2015) A dynamic programming algorithm for
leveraging probabilistic detection of energy theft in smart home. IEEE Trans Emerg Topics
Comput 3(4):502–513
10. Neto EAA, Coelho J (2013) Probabilistic methodology for technical and non-technical losses
estimation in distribution system. Electric Power Syst Res 97:93–99
11. Sahoo S, Nikovski D, Muso T, Tsuru K (2015) Electricity theft detection using smart meter
data. In Innovative smart grid technologies conference (ISGT), 2015 IEEE power & energy
society, February, pp 1–5. IEEE
12. Tangsunantham N, Ngamchuen S, Nontaboot V, Thepphaeng S, Pirak C (2013) Experimental
performance analysis of current bypass anti-tampering in smart energy meters. In: Telecom-
munication networks and applications conference (ATNAC), 2013 Australasian, November,
pp 124–129. IEEE
13. Dineshkumar K, Ramanathan P, Ramasamy S (2015) Development of ARM processor based
electricity theft control system using GSM network. In: Circuit, power and computing
technologies (ICCPCT), 2015 international conference on, March, pp 1–6. IEEE
14. McLaughlin S, Holbert B, Zonouz S, Berthier R (2012) AMIDS: a multi-sensor energy theft
detection framework for advanced metering infrastructures. In: Smart grid communications
(SmartGridComm), 2012 IEEE third international conference on, November, pp 354–359.
IEEE
15. Depuru SSSR, Wang L, Devabhaktuni V (2011) Electricity theft: overview, issues, prevention
and a smart meter based approach to control theft. Energy Policy 39(2):1007–1015
16. Prasad J, Samikannu R (2017) Overview, issues and prevention of energy theft in smart grids
and virtual power plants in Indian context. Energy Policy 110:365–374
17. Guerrero JI, Monedero Í, Biscarri F, Biscarri J, Millán R, León C (2014) Detection of non-
technical losses: the project MIDAS. Advances in secure computing, Internet services, and
applications, pp 140–164
18. Bula I, Hoxha V, Shala M, Hajrizi E (2016) Minimizing non-technical losses with point-to-point
measurement of voltage drop between “SMART” meters. IFAC-PapersOnLine 49(29):206–211
19. Saha M, Thakur S, Singh A, Agarwal Y (2014) EnergyLens: combining smartphones with
electricity meter for accurate activity detection and user annotation. In: Proceedings of the 5th
international conference on Future energy systems, June, pp 289–300. ACM
20. Han W, Xiao Y (2016) FNFD: a fast scheme to detect and verify non-technical loss fraud in
smart grid. In: Proceedings of the 2016 ACM international on workshop on traffic measurements
for cybersecurity, May, pp 24–34. ACM
Chapter 3
Renewable Energy Conversion:
Sustainable Energy Development
and Efficiency Enhancement of Solar
Panels: A Review
Bhavik J. Pandya, Megha C. Karia, and Kamlesh B. Sangani
1 Introduction
In the present time, Solar PV is rapidly growing to become a top source of electric
power generation. Worldwide installed capacity is the third-largest source of renew-
able resource after hydro as well as wind. It should be noticed that the progress in
the development in the area of solar thermal is quite slower, but it indicates that it is
in just the beginning period and it may be a major source for electricity generation
later in century. Solar power is the most abundant form of energy in our mother earth
and it is a mandatory source for its inhabitants. Now this paper presents an overview
on generation of electricity from solar energy utilization. This can be achieved most
simply by exploiting the heat contained in the sun’s radiation.
1.1 Present Scenario in Energy Generation
The power generation capacity of installed power plants in India as on 29.02.2012

is 1,90,593 MW (share of renewable energy sources is 22,253 MW). The gross
electricity generated in the year of 2011–12 in India (up to February 2012) includes
B. J. Pandya (B) · M. C. Karia

V.V.P. Engineering College, Rajkot, India
e-mail: bhavikpandya.education@gmail.com
M. C. Karia
e-mail: megha.karia.me@vvpedulink.ac.in
K. B. Sangani
Super Specialist Technocrats LLP, Rajkot, India
e-mail: kamleshsanganisir@gmail.com

https://doi.org/10.1007/978-981-15-8820-4_3
26 B. J. Pandya et al.
bring from Bhutan was 798.9 Billion Units. The energy requirement, availability,
and shortage in India during April 2011–February 2012 are given below:
Year Energy requirement Energy availability Deficit (MU) Deficit (%)

(MU) (MU)
2011–12 (till Feb 2012) 853,324 782,124 71,200 8.3%
1.2 History of Electricity Generation from Solar Energy

Resources:
Solar thermal power exploitation, the solar heat utilization can be found by going
back in history to Archimedes but it must be included that the generation of power by
the utilization of solar energy is recent the use of the sun as a source of heat, can be
traced back at least to Archimedes but its application as a means to generate power is
more recent. The utilization of solar heat is started from nineteenth century; parabolic
reflectors were used to concentrate the heat of sun and due to this heat steam produce
which used to drive steam engine. From the beginning of twentieth century, solar
energy was applied to get power in engine for work like pump water for irrigation
in the agricultural sector; this technique was used in Egypt. The first power plant
operated by solar thermal power was established in 1960s in Italy. There is another
method for generating electricity from solar with the help of an electronic device
which is known as photovoltaic or PV cell. Another method of transformation of
sunlight to electricity discovered, by the research of scientist named Antoine-Ce’sar
Becquerel. The first person who observed the effect of photovoltaic conversion to the
voltage when sun rays supplies to electrode was Becquerel. After this, a gold-coated
selenium solar cell was invented by Charles Fritts. This first cell was inefficient
but it gave thought for large scale electricity generation by solar. In 1941, the first
silicon solar cell was invented by Russell Ohl after a number of experiments. After
the 13 years an American scientist Gerald Pearson, Calvin Fuller, and Daryl Chapin
developed a cell which has conversion efficiency of 6%.
1.3 Potential of Solar Energy
We all know that sun has energy which produces by nuclear reactions at its center.
The electromagnetic radiation helps it to reach till earth’s surface. This radiation is
passed from 7% ultraviolet zone, 36% visible radiation zone, and 56% infrared zone
of electromagnetic spectrum.
3 Renewable Energy Conversion: Sustainable Energy Development … 27
Fig. 1 Global power generation in 2013 and 2030 [1]
1.4 Solar Energy and the Earth
Solar constant at the distance of the earth from the sun 1367 W/m2
Total solar energy reaching the earth in a year 3,400,000 EJ
Total solar flux reaching earth 1.083108 GW
Average solar energy density at earth’s surface 170 W/m2
Source World Energy Council
1.5 Current Situation of Solar Power Plant in India
In January 2016, a new international body for solar energy utilization, International
solar alliance (ISA) has been initiated by the hardly efforts of Indian Prime minister
Narendra Modi and with the support of French president Francois Hollande and
also established it’s headquarter at Gwal-Pahari, Gurgaon. The development and
utilization of solar energy and products works by it for countries between Tropic of
cancer and Tropic of Capricorn is the main focus of ISA. More than 120 countries
agreed and sign the alliance at the summit held at Paris named Paris COP21 climate
change (Figs. 1 and 2).
In India utilization of solar PV is at peak so some statistical data is shown below
(Table 1).
Installed Solar PV till 31st March 2019
Fig. 2 Global horizontal irradiance (Source NREL)
2 Classification of Solar Power Plant
There are certain methods for conversion of energy of sun to the electrical power.
The first option is the direct conversion to electricity mainly by the following
methods:
Table 1 Installed solar PV

Year Cumulative capacity (MW)
till 31st March 2019 (Source
Wikipedia) 2010 161
2011 461
2012 1205
2013 2319
2014 2632
2015 3744
2016 6763
2017 12,289
2018 21,651
2019 28,181
1. Photovoltaic.
2. Photogalvanic.
3. Photoemissive.
4. Photomagnetic.
The direct conversion is practically not adopted for commercial purpose for large
power demand due to high cost of cell.
Thermal energy obtained by solar energy can also be directly converted into
electricity by the following methods:
1. Thermo electric.
2. Thermionic.
3. Ferro electric.
4. Magneto hydrodynamic.
5. Electro gas dynamic.
Above-mentioned methods have generally very low rate of conversion of solar
energy to electricity, so this power system is not cost effective.
There is yet another very important way for electrical conversion from solar this
is thermodynamic way, in which solar heat is converted into thermal energy and this
thermal energy will be converted to shaft work through heat engine based on the
principle of either Rankine cycle, Stirling cycle, or Brayton cycle and shaft work
(mechanical energy) into electricity using alternator.
2.1 Concentrating Solar Power (CSP)
In this system, solar rays are concentrated by mirrors and lenses and then it reflects
area under sunlight to small section beam, and thus by the help of this concentrated
solar beam we can utilize this solar energy. Here lenses capture large area and utilize
its heat in small area so we can get efficient work output. Here sun rays are focused
Parabolic trough system Parabolic dish system Solar power tower
Fig. 3 [2]
on working medium, then working medium gets heat by concentrated solar rays,
thus power generation takes place. It is mandatory to mention that for the purpose
of focusing sun rays various tracking techniques are used. Parabolic trough, solar
power tower, sterling dish, linear Fresnel reflector, and many more technologies exist
in CSP and some in developing stage also (Fig. 3).
CSP applicaons:
T<100ºC 100ºC<T<300ºC 300ºC<T<600ºC T>600ºC
gas reforming
solar water heater, chemical,steam and
food and texle industry, steam
solar dryer, food metal producon,
industries, producon, chemical
process industries electricity producon
industries
parabolic dish,
parabolic trough parabolic trough central tower central tower
Layout 1 Application of concentrated solar power system application

poly
crystalline
crystalline mono
Si crystalline
Dye-
GaAs
Sensized
HIT
Hybrid PV
perovkite
single
solar panels juncon
double
a-Si
juncon
tripple
thin film CdTe/CdS
juncon
CIS/CIGS
single layer
organic
multy layer
Layout 2 Different solar PV technologies
2.2 Concentrated Solar Power System Application
2.3 Solar Photovoltaic Cell Power Generation
The photovoltaic cell is solid-state device composed of thin layers of semiconductor

materials which produce an electric current when light falls on it. It contains number
of cell filled by photovoltaic material. Mono-crystalline Si, Poly-crystalline Si, amor-
phous Si, Cu-indium sulfide, and Cd-telluride are materials which are currently used
for PV. Generally Poly-crystalline Si and Mono-crystalline Si are extensively used

in India. Silicon gives optimum result so it is extremely suitable for PV applications
among all the materials available in the earth [18].
2.4 Different Solar PV Technologies
2.5 Classification of Solar PV Cells
I. Crystalline Si PV cell
A. Gallium Arsenide (GaAs)
GaAs solar cell is part of a compound semiconductor which are a combination
of gallium (Ga) and arsenic (As) having the same structure as silicon [7]. They
have higher efficiency and weight is lower compared to silicon produced PV
cells. Here it should be mandatory to mention that economically GaAs is much
costly then mono- and poly-crystalline Si cell.
B. Mono-crystalline PV cell
C. Poly-crystalline PV cell
Note: B & C we will discuss in detail in later.
D. Emitter Wrap-Through cell (EWT) cells
EWT utilizes its full cross-sectional area so it can focus large amount of solar
radiation. Reason behind this is it has modified design of back content cell with
a front collection junctions [8]. The conversion efficiency of EWT is between
15 and 20%.
E. Edge-defined film-fed growth (EFG) cells
For the reduction of cost of wafers EFG Si ribbon is used. The kerf losses
(which generated because of wafer sawing and ingot) elimination and cutting
down of poly-Si consumption is a major benefit of EFG. As per the current
research data, researchers found 16% conversion efficiency of two materials
when reduced shunting.
II. Thin-film PV cell
Thin-film solar cells generally made by very thin layers (<10 µm) settled on the
metals, polymers, or glass, by sputtering process [6]. The detailed classification
of thin-film PV is given below.
A. Amorphous silicon (a-Si) solar cells
Amorphous silicon solar cells have a disordered structure form of Si and have
40 times higher light absorption rate compared to the mono-Si cells. They are
majorly utilized and most developed thin-film solar cells. At laboratory level,
efficiency of single junction a-Si cells may reach up to 12.2%.
B. Cadmium Telluride (CdTe) or CdS cells
Cadmium Telluride thin-film PV solar cells consist mainly two types of raw
materials which are cadmium and tellurium, where cadmium is a by-product
of zinc mining whereas tellurium is found out by copper processing. Both of
them have lower production cost and higher cell efficiency (higher than15%)
compare to the a-Si cells. Here highest 21% conversion efficiency is achieve by
a leading company named “First solar” in august 2014.
C. Copper-Indium-Selenide (CIS) and Copper-Indium-Gallium-Diselenide
(CIGS) cells
CIS cells are competing with other thin-film cells due to their high conver-
sion efficiency 21% reported at laboratory level. Here interfacial, stable back
contact, grain boundry, and junction activation are the main problems with
CIGS. Bareness of indium is also one of the major problem in CIGS.
III. Organic PV cell
It is developing and latest technology for alternate material of solar cell. Dispos-
able, flexible, light weight, less production cost, semitransparent, and low
material required are a characteristics of it [6, 9, 10] (Table 2).
Table 2 Laboratory-based efficiency of terrestrial PV cell and sub-module measured at STC [11]
Classification Efficiency Area Voc Jsc Fill Description
% Cm2 (Open (Current factor
circuit density) %
voltage)V MA/cm2
Silicon
Si (crystalline) 25.6 ± 0.5 143.7 0.740 41.8 82.7 Panasonic
(da) HIT, rear
Junction
Si (multi crystalline) 21.25 ± 0.4 242.74 0.6678 39.80 80.0 Trina solar
(t)
Si (thin transfer sub-module) 21.2 ± 0.4 239.7 0.687 38.50 80.3 Solexel
(ap)
Si (thin-film minimodule) 10.5 ± 0.3 94.0 0.492 29.7 72.1 CSG solar
(ap) (<2 µm on
glass; 20
cells)
III-V cells
GaAs (thin-film) 28.8 ± 0.9 0.9927 1.122 29.68 86.5 Alta devices
(ap)
GaAs (multicrystalline) 18.4 ± 0.5 4.011 0.994 23.2 79.7 RTI,Ge
(t) substrate
InP (crystalline) 22.1 ± 0.7 4.02 (t) 0.878 29.5 85.4 Spire,
epitaxial
Thin-film chalcogenide
CIGS (minimodule) 18.7 ± 0.6 15.892 0.701a 35.29 75.6 Solibro, 4
(da) serial cells
CdTe (cell) 21.0 ± 0.4 1.0623 0.8759 30.25 79.4 First solar
(ap) on glass
(continued)
Table 2 (continued)
Classification Efficiency Area Voc Jsc Fill Description
% Cm2 (Open (Current factor
circuit density) %
voltage)V MA/cm2
Amorphous/microcrystalline
Si
Si (amorphous) 10.2 ± 0.3 1.001 0.896 16.36 69.8 AIST
(da)
Si (microcrystalline 11.8 ± 0.3 1.044 0.548 29.39 73.1 AIST
(da)
Organic
Organic thin-film 11.0 ± 0.3 0.993 0.793 19.40 71.4 Toshiba
(da)
Organic (minimodule) 9.7 ± 0.3 26.14 0.806 16.47 73.2 Toshiba (8
(da) series cells)
Perovskite
Perovskite thin-film 15.6 ± 0.6 1.020 1.074 19.29 75.1 NIMS
(da)
3 Comparison of Various Solar PV Cell
3.1 Laboratory-Based Efficiency of Terrestrial PV Cell

and Sub-Module Measured at STC
3.2 Laboratory-Based Efficiency of Terrestrial PV Modules

Measured at STC
From the above laboratory experimental data, we have seen the performance of
various PV cells but generally in India only two types of PV cells are used, i.e., (1)
Mono-crystalline Si solar PV cell and (2) Poly-crystalline Si solar PV cell (Table 3
and Fig. 4).
Mono-Crystalline Solar PV Cell
Mono-Si photovoltaic panels are the most efficient panels. Each cell is made by
only one Si crystal. Its efficiency is also much higher than the Poly-Si and thin-
film. As per economic point, it is expensive than both Poly-Si and thin-film [1] the
method used for production of mono-crystalline Si is known as Czochralski process.
In Czochralski method is done in cylindrical boules. After this boules are cut into
thin “pseudo-square” shapes [5]. To manufacture this PV cell, diffusion, chemical
etching, antireflection coating, edge isolation, and development of metal this process
Table 3 Laboratory-based efficiency of terrestrial PV modules measured at STC [11]

Classification Efficiency Area Voc Jsc Fill Test Description
% (cm2 ) V A factor center
Si (crystalline) 22.9 ± 0.6 778 (da) 5.60 3.97 80.3 Sandia UNSW/Gochermann
(9/96)
Si (large 22.8 ± 0.6 15,738.9 69.36 6.459 80.0 NREL Sun power (96 serial
crystalline) (ap) (6/15) cells)
Si 19.2 ± 0.4 15,126.5 77.93 4.726 78.93 FhG-ISE Trina solar (120
(multicrystalline) (ap) (6/15) serial cells)
GaAs (thin-film) 24.1 ± 1.0 858.5 10.89 2.255 84.2 NREL Alta devices
(ap) (11/12)
CdTe (thin-film) 18.6 ± 0.6 7038.8 110.6 1.533 74.2 NREL First solar,
(ap) (4/15) monolithic
CIGS (Cd free) 17.5 ± 0.5 808 (da) 47.6 0.408 72.8 AIST Solar frontier (70
(6/14) cells)
CIGS (thin-film) 15.7 ± 0.5 9703 28.24 7.254 72.5 NREL Miasole
(ap) (11/10)
a-Si/nc-Si 12.3 ± 0.3 14,322 280.1 0.902 69.9 ESTI TEL solar, Trubbach
(tandem) (t) (9/14) Labs
Organic 8.7 ± 0.3 802 (da) 17.47 0.569 70.4 AIST Toshiba
(5/14)
Fig. 4 Comparison between laboratories-based PV cell and module efficiencies
done on pseudo-square shape wafers [4]. Generally, mono-crystalline Si has 15–20%

infield conversion efficiency [3, 5]. Lower cost silicon tri-Si and poly-Si are used as
an alternative of it because of its higher cost.
Poly-Crystalline Solar PV Cell

Here number of small silicon crystal is used as raw material. To manufacture
this, large amount of molten Si is solidified and converted into orient crystals. By
producing these crystals in same direction, it makes cast square shape ingots of it,
which will then cut to blocks to wafers [6]. Poly-Si has lower efficiency than mono-Si
but its cost is also low so generally it is used for module manufacturing. Currently,
it is noticed that on large scale filed conversion efficiency of Poly-Si is from 13 to
16% [5]. Recently it is found that conversion efficiency of mono-Si and poly-Si is
same for some products like SUNTECH 250 W. SUNTECH 250 W has a fill factor
and conversion efficiency of 77.52% and 15.38%, respectively [7].
4 Comparison of Mono-Crystalline and Poly-Crystalline

Solar Cell by AHP
AHP means Analytical Hierarchy Process.

In this process, we consider some parameters like mechanical properties,
electrical, environment, customer satisfaction, and economic criteria. Here we
consider above-mentioned parameters analyses various kind of solar panels.
According to that we do comparative analysis of that solar panels and come to
conclude best brand for solar panels as per AHP. Here the evaluation is done for
200 W capacity solar panels of different manufacturer companies and find best among
all [12].
Environment P1 P2 P3 P4 P5 P6
Area 1.40 1.27 1.46 1.27 1.48 1.16
Material Poly Mono Mono Mono Poly Mono
Electrical characteristics
PTC power rating (W) 175 180 179 184.8 177.5 185.9
STC power per unit of area (W/m2 ) 142.1 156.7 136.5 156.7 135.1 172.3
Peak efficiency (%) 14.21 15.67 14.2 15.67 13.5 17.2
Power tolerances (%) −9/+9 0/+3 −3/+32 0/+5 −3/+3 0/+10
Number of cells 50 72 54 72 54 96
Imp (A) 8.16 5.42 7.89 5.17 7.60 5.59
Vmp (V) 24.50 36.90 25.38 38.70 26.30 55.80
Isc (A) 8.70 5.80 8.24 5.50 8.22 3.83
Voc (V) 30.80 45.60 33.53 45.90 33.30 68.70
NOCT (°C) – 45 – 45 – –
Temp. coefficient of power (%K) −0.50 −0.40 −0.50 −0.38 −0.45 −0.29
Temp. coefficient of voltage −0.5 −0.4 −0.5 −0.38 −0.45 −0.17
(V/K)
Series fuse rating (A) 15 10 15 15 15 15
Maximum system voltage (V) 600 1000 600 10,000 600 600
Lower energy density(W/m2 ) 11.52 11.92 11.82 13.28 11.26 14.89
Mechanical characteristics
Frame color Bronze Clear Black Clear Clear Black
(continued)
(continued)
Environment P1 P2 P3 P4 P5 P6
Weight (kg) 15.4 14.5 35 15.5 18 15
Financial properties
Price ($) 300 300 499 300 319 600
Cost per Watt ($) 1.05 1.05 1.75 1.05 1.05 2.10
Customer satisfaction
Service support 3 1 6 5 2 4
Spare part 5 2 6 4 1 3
Reliability 5 2 6 4 3 1
The above table which shows AHP analysis by considering environmental,

mechanical, electrical, customer satisfaction, and financial criteria. After studying
this table we came to conclude that solar panel brand P6 is the best suitable. Here it
must be mention that these results are for specific cases and for a particular location,
it may be changed as per change in locations [12].
4.1 Solar Tracking System
The solar trackers are generally used for minimizing the shading effect and for
performance enhancement of solar PV.
The solar tracker can be classified as (i) fixed-rack tracker; (ii) single-axis tracker;
and (iii) double-axis tracker.
(1) Fixed-rack tracker: As the name suggests, it cannot change its angle as per sun’s
movement. It generally remains fixed at particular angle. At the same time its cost is
least among all three because it does not require any additional device. It is durable
as well as less maintenance required.
(2) Single-axis tracker: Single-axis tracker can just track one daily motion of sun
either north-south or east-west. It can rotate in only single axis of rotation either
horizontal or vertical. It can generate higher amount of electricity compare to fixed
tracker but it also requires extra device for the tracking purpose. Due to this extra
device its cost is higher.
(3) Double-axis tracker: It can track in both north-south as well as east-west motion
of sun. As name indicates it has two-axis rotation in horizontal and vertical direction.
So it has highest electricity production capacity but its cost is also highest among
the three.
1. Fixed rack 2. Single axis tracker 3. Double axis tracker

Tracker [26]
4.2 Novel and Emerging Solar Cell Technology Concept
In the above study, we just discussed few commercially available traditional solar
cells. Carbon nanotubes, superlattice technology, quantum wells, hot carrier (HC),
and Quantum wires/dots are newly invented materials [13, 14]. Improve efficiency
and enhance performance is the main purpose for development of new technology.
Here nanomaterials and nanoscale components control band gap and improve
efficiency and performance.
A semiconductor which can absorb spectrum of particular size for the purpose of
match with solar spectrum is called Quantum dots. It has also one additional benefit
that it can generate multiple electron-hole pairs per photon. As mentioned above, for
efficiency improvement it can adjust band gap [14].
Generally conversion efficiency of carbon nanotubes varies from 3 to 4%. It is
generally coated with CdS—Cadmium sulfide and CdTe—cadmium telluride for
the purpose of trap photons and completely absorb it. It is noticed that it has good
mechanical and electrical properties.
4.3 Waste Solar Panels Recycling Technology
The recycling of end-of-life (EOL) panel is also one of the important parameters
which we generally neglect. There is a lack of knowledge and awareness regarding
this [15]. The table shown below is a comparison of various solar panels recycling
technology.
40
Key technology Method Result-Advantages Disadvantages Reference

Component repair Two kinds of methods to dismantle Can increase the output power of old Can be used only if the external [16]
and solar panels junction box and outer layer are aging
analyze faults in the junction box
Module separation Organic solvent method Mechanical pressure is critical to Organic liquid waste produced [17]
Compared to three methods: artificial suppress the swelling of the EVA. The Did not involve the resource recovery [18]
disassembly, use of organic solvent to silicon panel was recovered of Silicon [19]
dissolve components, and heat successfully with no damage Not sufficient for single component [20]
treatment The heat treatment process was the separation; still in the laboratory
Artificial disassembly, crushing, optimal solution research stage
cryogenic breaking, and electrostatic Obtained a mixture of different Organic liquid waste produced
separation types of materials Non-purified silicon wafers
Organic solvent-assisted ultrasonic significantly shortens the dissolution
method time of EVA in
Heat treatment and chemical etching organic solvents
methods Module separation
Recycling of silicon Cement-based thermal insulation Recycling of silicon Organic liquid waste produced [21]
system and chemical method
(continued)
B. J. Pandya et al.
(continued)
Key technology Method Result-Advantages Disadvantages Reference
Recycling of rare metals Grinding and hydrometallurgy Recycling of indium and gallium High price of chemicals [22]
Physical and chemical methods: Recycling of cadmium telluride; net High price of chemicals [23]
blasting, cost estimated at Process is complicated [24]
mechanical processing and $0.04–$0.06/W All types of mixtures, hard to separate [25]
dissolution of the semiconductor, and Recycling (CdTe and CIS)
precipitation, plating, and ion Mix of recycled poly-silicon,
exchange amorphous silicon, and CdTe
Recycling (CdTe and CIS) by wet solar panels; glass directly
mechanical recycle
treatment such as grinding and
flotation, or dry mechanical
processing methods such as vacuum
blasting
Treated multiple types of solar panels
together. Two physical methods were
used: panels were broken and then
heat-treated or broken with a hammer
3 Renewable Energy Conversion: Sustainable Energy Development …
41
5 Conclusion
After exhaustive review based on available literature presented by renowned

researchers in field of solar power generation using photo volatile technology is
discussed briefly here. We have observed various characteristics of solar PV cell and
effect of variables in solar PV cells.
Generally electrical performance of PV cells and modules are measured by manu-
facturer at standard temperature condition but in actual power plant modules operate
at different climatic conditions. Hence there is variation in performance, conversion
efficiency, and generated power are diverse from experimental results within labo-
ratory and real field environment. A noticeable deviation observed for experimental
result between solar PV cell and module.
Here we have mentioned laboratory-based performance analysis of different PV
cells like silicon crystalline, thin-film, organic, etc. and laboratory-based performance
analysis of different PV modules; done by various researchers. After comparing it,
it is observed that there is less power generated in PV module.
In India, poly-crystalline and mono-crystalline PV panels are majorly used so
here we have discussed a comparative AHP (Analytical Hierarchy Process) analysis
performed by honorable researchers and made summary of it as per different condi-
tions. At the end of AHP analysis, we came to the conclusion that mono-crystalline
panel of brand P6 is best suitable according to financial, mechanical, environmental,
electrical, and customer satisfaction criteria.
Here we also discussed novel solar cell technologies and some methods for
recycling waste solar panels.
References
1. Manishkumar, Arun kumar, Performance assessment and degradation analysis of solar

photovoltaictechnologies: a review
2. Sharma A (2011) Comprehensive study of solar power in India and world. India. Renew Sustain
Energy Rev 15: 1767–1776
3. Park S, Pandey A, Tyagi V, Tyagi S (2014) Energy and exergy analysis of typical renewable
energy systems. Renew Sustain Energy Rev 30:105–123
4. Maehlum MA (2013) Which solar panel type Is best? Mono vs. polycrystalline vs. thin–film.
Energy Inf
5. Tyagi V, Rahim NA, Rahim N, Jeyraj A, Selvaraj L. Progress in solar PV technology: research
and achievement. Renew Sustain Energy Rev
6. El Chaar L, El Zein N (2011) Review of photovoltaic technologies. Renew Sustain Energy Rev
15:2165–2175
7. Akinyele D, Rayudu R, Nair N (2015) Global progress in photovoltaic technologies and the
scenario of development of solar panel plant and module performance estimation− Application
in Nigeria. Renew Sustain Energy Rev 48:112–139
8. Gee JM, Schubert WK, Basore PA (1993) Emitter wrap-through solar cell. In: Photovoltaic
specialists conference, conference record of the twenty third IEEE, pp 265–270. IEEE
9. Pandey A, Tyagi V, Jeyraj A, Selvaraj L, Rahim N, Tyagi S (2016a) Recent advances insolar
photovoltaic systems for emerging trends and advanced applications. Renew Sustain Energy
Rev 53:859–884
10. Goetzberger A, Hebling C, Schock H-W (2003) Photovoltaic materials, history, status and
outlook. Mater Sci Eng: R: Rep 40:1–46
11. Green MA, Emery K, Hishikawa Y, Warta W, Dunlop ED (2016) Solar cell efficiency tables
(version 47). Progress Photovolt: Res Appl 24
12. The selection of the best solar panel for the photovoltaic systemdesign by using AHP
FigenBaloa, LütfüŞağbanşuab, * Energy Procedia 100 (2016) 50–53
13. Pandey A, Tyagi V, Jeyraj A, Selvaraj L, Rahim N, Tyagi S (2016b) Recent advances in solar
photovoltaic systems for emerging trends and advanced applications. Renew Sustain Energy
Rev 53:859–884
14. Jha AR (2008) MEMS and nanotechnology-based sensors and devices for communications,
medical and aerospace applications. CRC Press
15. Xu Y, Li J, Tan Q, Lauren Peters A (2018) Global status of recycling waste solar panels: a
review. ChinaWaste Manag
16. Lin W, Chen E, Sun YL (2011) Analysis of old photovoltaic component junction box
disassembling mode. Sol Energy 7:26–29
17. Doi T, Tsuda I, Unagida H, Murata A, Sakuta K, Kurokawa K (2001) Experimental study on
PV module recycling with organic solvent method. Sol Energy Mater Sol Panels 67:397–403
18. Dong L (2009) Research on waste crystalline silicon solar panels resource recovery
19. Kim Y, Lee J (2012) Dissolution of ethylene vinyl acetate in crystalline silicon PV modules
using ultrasonic irradiation and organic solvent. Sol Energy Mater Sol Panels 98:317–322
20. Klugmann-Radziemska E, Ostrowski P (2009) Chemical treatment of crystalline silicon solar
panels as a method of recovering pure silicon from photovoltaic modules. Renew Energy
2009:1–9
21. Fernandez LJ, Ferrer R, Aponte DF, Fernandez P (2011) Recycling silicon solar cell waste in
cement-based systems. Sol Energ Mat Sol C 95:1701–1706
22. Guangdong Xiandao Rare Material Co. Ltd (2011) Recovery of copper indium gallium selenide
thin-film solar panel, involves crushing solar panel, soaking in sulfuric acid, filtering, extracting,
separating, stripping extraction liquid, adding reducing agent to raffinate and filtering (Chn)
CN 103184338-A[P] 2011-12-29
23. Berger W, Simon FG, Weimann K, Alsema EA (2010) A novel approach for the recycling of
thin film photovoltaic modules. Resour Conserv Recy 54:711–718
24. Granataa G, Pagnanelli F, Moscardini E, Havlik T, Toro L (2014) Recycling of photovoltaic
panels by physical operations. Sol Energy Mater Sol Panels 123:239–248
25. Kang H, Hong T∗, Jung S, Lee M. Techno-economic performance analysis of the smart solar
photovoltaic blinds considering the photovoltaic panel type and the solar tracking method
26. Hafez AZ, Yousef AM, Harag NM (2018) Solar tracking systems: technologies and trackers
drive types –A review. Renew Sustain Energy Rev 91:754–782. https://doi.org/10.1016/j.rser.
2018.03.094
27. Heslop S, MacGill I (2014) Comparative analysis of the variability of fixed and tracking
photovoltaic systems. Sol Energy 107:351–364. https://doi.org/10.1016/j.solener.2014.05.015
28. König D, Casalenuovo K, Takeda Y, Conibeer G, Guillemoles J, Patterson R et al (2010)
Hot carrier solar cells: principles, materials and design. Phys E: Low-Dimens Syst Nanostruct
42:2862–2866
Chapter 4
Evaluation of Thermal Degradation
Behavior of Cardboard Waste
Samit Kumar Singh and Sadanand A. Namjoshi
1 Introduction
The waste-to-energy is really a challenging task and to accomplish this task, the
pyrolysis is one of the best options. The composition of MSW varies because of socio-
economical status of the society, economic development, demographic division, and
waste collection potential also for waste-to-energy options. The extensive rises in the
MSW generation having severe impact on environment but its disposal associated
with social and economic problems.
The pyrolysis having gas, char, and liquid yield products; it means three types of
fuels are available. The significance of pyrolysis is also due to the decomposition
of polymers into fuels which are not easily degradable and contaminate atmosphere
heavily when thrown as a waste at open places. The design of pyrolyzer is based
on kinetics studies which correlate the weight degradation with respect to time and
temperature; it also represents space or in other words, volume requirement of fuel
in the form of pyrolytic products with respect to time temperature. The quantities of
yield products are also depends on available space otherwise temperature gradient
will exist and conduction losses increase.
Garcia et al. [1] focused kinetics of pyrolysis at different heating rate between
1.5 and 200 °C min−l . After observing the collected results, a theoretical study was
carried out in connection with the effect of heating rate and the transmission of heat
on the kinetic parameters which can be obtained using the experimental data.
Fontet al. [2] investigated the kinetics of the almond shells impregnated and non-
impregnated with CoCl2 , polyethylene lignin, and MSW with the help of an analytical
S. K. Singh (B) · S. A. Namjoshi

Madhav University, Pindwara (Sirohi) 307026, Rajasthan, India
e-mail: samit.meet@gmail.com
S. A. Namjoshi
e-mail: namjoshisadanand@yahoo.co.in

https://doi.org/10.1007/978-981-15-8820-4_4
46 S. K. Singh and S. A. Namjoshi
pyro probe 1000 or pyroprobe 100, thermogravimetric fluidized sand bed reactor was
used to understand the thermal decomposition of heterogeneous materials.
Wu et al. [3] reported kinetic study of painted printed and writing paper by TGA
in nitrogen atmosphere for the temperature range is from 450–900 K at 1, 2, and
5 K/min heating rates with two stages of decomposition.
Park et al. [4] suggested a new method for estimation of kinetics constants of
low-density polyethylene (LDPE), linear low-density polyethylene (LLDPE), and
high-density polyethylene (HDPE), at heating rates of 100 °C/min and 500 °C/min.
The numerous analytic methods mentioned were used for comparing with the kinetic
parameters estimated from the dynamic method.
Sorum et al. [5] investigated in-depth knowledge of pyrolysis process and estima-
tion of chemical kinetics important constituents of MSW. TGA estimates the kinetic
constants which are executed at 100 °C/min heating rate in N2 gas atmosphere. The
degradation of hemicellulose, cellulose, and lignin was shown by three autonomous
parallel reactions the cellulosic fraction of MSW.
David et al. [6] explained pyrolysis process depending upon the numbers of TGA
experimental work on cardboard. The research work includes the determination of
kinetics parameters for cardboard. These research works also focus pyrolysis process
with the support of numbers of TGA experimental works also compare the results
of graphical method with numerical results which are in good level of satisfaction.
Molto et al. [7] focused on kinetic studies for the decomposition of used cotton
fabrics and for explaining the behavior of decomposition all the runs were performed
and have been proposed and tested. In the pyrolysis of the used cotton fabric waste,
the model comprises two parallel reactions during decomposition.
Miranda et al. [8] studied thermal degradation behavior of textile waste by TGA
at different heating rates also decomposition was studied by semi-batch pyrolysis
process.
The DTG curves show three peaks; two out of which were hemicellulose and
cellulose and one might be due to decomposition synthetic polymer. The major
research in present work focuses on thermal degradation behavior of cardboard waste
at 5, 10, and 15 °C/min heating rate and also ash obtained during decomposition of
cardboard analyzed using Scanning Electron Microscope (SEM).
2 Experimental Methodology
2.1 Material
The experiments were conducted on cardboard collected from waste packing boxes
which was sheared into small pieces of 1–2 mm by using scissors. The size of the
particle used in TGA having micro-weight and sample amount limitation of container.
Table 1 indicates proximate, ultimate result, and HHV of cardboard.
4 Evaluation of Thermal Degradation Behavior of Cardboard Waste 47
Table 1 Proximate analysis, ultimate analysis, and higher heating value cardboard
Fixed Volatile Ash Carbon Hydrogen Nitrogen Sulfur Oxygen HHV
carbon matter (% (% wt) (% wt) (% wt) (% wt) (% wt) MJ/kg
(% wt) (% wt) wt)
12.6 74.9 12.4 42.13 5.3 0.18 0.54 39.51 15.35
2.2 Thermogravimetric Analysis (TGA)
The engineering design of modern pyrolysis reactors and gasifiers requires a deep
knowledge of kinetics studies for different MSW constituents at different temperature
ranges and heating rates. The TGA and DTG were performed in TG/DTG system.
Samples’ temperature range varies from room temperature (RT) to 600 °C with 5,
10, and 15 °C/min heating rate in nitrogen atmosphere and flow rate of gas was
50 ml/min. The sample size of cardboard was taken as 6 mg. Samples were placed
in platinum container and reference used was α alumina.
3 Kinetics
It is necessary to understand the reaction mechanism of pyrolysis for obtaining the

reactions with respect to its kinetic parameters. The heated deterioration of MSW
constituents is explained to be proceeding through many complicated reactions
[5, 6, 8].
A method was formulated to calculate activation energy from single dynamic
TGA data employing following model [10, 11].
Here F062 is heating rate, E is activation energy A is pre-exponential or frequency
factor, R is universal gas constant (8.314 kJ mol−1 K−1 ).
This equation has no exact solution; hence, different approximations have been
done by Flynn wall Ozawa method [12].
For cardboard decomposition, in case TGA, the Flynn wall Ozawa [12] method
used for activation energy for three heating rates are used.
4 Results and Discussion
Figure 1 represents Thermogravimetric curves of cardboard at 5, 10, and 15 0 C/min

heating rates, respectively, while Fig. 2 represents the corresponding Differential
Thermogravimetric curves for same heating rates.
The decomposition is observed in three stages as shown in Figs. 1 and 2. In first
stage, moisture present in sample gets removed and hemicellulose gets decomposed.
The degradation rate in first stage is slow due to low temperature range and mass
100.00
80.00 5 C/min
10 C/min
TG (%wt)
60.00
40.00 15 C/min
20.00
0.00
0 100 200 300 400 500 600
Temperature (° C)
Fig. 1 TG curves
500
450 5 C/min
400
350
10 C/min
300
mg/min
250
200 15 C/min
150
100
50
0
0 100 200 300 400 500 600
Temperature ( ° C)
Fig. 2 DTG curves
transfer resistance [1, 5, 6, 8]. The degradation of cellulose and lignin occurs in
second and third stage, respectively. The similar results are reported by Sorum et al.
[5], David et al. [6], Miranda et al. [8] and which substantiates the present work.
Figure 1 represents same TG curves pattern for heating rates 5, 10, and 15 °C/min;
however, rates and kinetics parameters are differing. During second stage, degrada-
tion occurred is faster because cellulose gets decomposed compared to first and third
stage where hemicellulose and lignin are decomposed, respectively, as heating rate
increases which is represented by TG curves in Fig. 1 [5]. The loss in weight escalates
in all three phases at a quicker amount as heating rate rises which is quite obvious that
heating rate enhances the values of kinetics parameters too. In connection to three
stages of decomposition on the TG curves correspondingly, three peaks are noticed
on the DTG curves. DTG curves show as heating rates expand there will be a lateral
shift toward greater temperatures and which may be the united effects of the heat
transfer at various heating rates and resulting in delayed the decomposition as stated
by Sorum et al. [5]. Figure 1 reports that rise in heating rate gives the greater value
of activation energy, which can be attributed to the formation of active molecules
also enhances in molecular collision at high heating rates [5, 8]. It also shows that
the activation energy value is maximum in second stage in comparison to other two
stages in all cases all heating rates. During third phase of decomposition, activa-
tion energy value gets lowered because of low deterioration rate. It is observed that
amount of decaying rises with the rise in heating rate and a declination is observed in
total decaying. It is because of the inadequate decaying time for concluding reaction
with rising heating rates [8]. In case of all cellulosic fractions, the major weight loss
occurs between the temperatures ranges of 250–400 °C as also reported by Sorum
et al. [5].
Figure 2 represents the DTG curves which are nothing but decomposition rate with
respect to temperature in association with 5, 10, and 15 °C/min heating rates. Results
show that rise in heating rate leads to rise in weight loss and corresponding maximum
temperature also increases linearly due to delay in decaying which penetrates extra
heat into the specimen for shorter while [5]. In case of cardboard, the three peaks
are commonly observed and the temperature range in which maximum peak can
be observed varies from 175 to 250 °C, 300 to 425 °C, and 450 to 525 °C which
corresponding to hemicellulose, cellulose, and lignin, respectively [5, 6].
The description of less marked shoulder on the DTG curves for hemicellulose
in case of cardboard is due to the minor content of hemicellulose and/or catalytic
effects. Catalytic effects generated by inorganic species such as ash and in the case
of cardboard, residues from sulfate process causes the decaying of cellulose to occur
at reduced temperature [5].
Tables 2 and 3 present the values of decomposition with temperature range and
kinetic constants for different stages of decomposition, respectively.
Table 2 % Weight loss in stages of decomposition corresponding to temperature range

Heating Stage I Stage II Stage III %
Rate % Loss Range of % Loss Range of % Loss Range of Residual
(°C/min) in temp °C in temp °C in temp °C remains
weight weight weight
Cardboard waste
5 6 30–193.5 52.2 193.5–349 27.2 349–545 14.5
10 6 30–193.3 53 193.3–366.5 24 366.5–539.6 17
15 7 30–209.5 50.3 209.5–353.1 27 353.1–532.9 15.7
Table 3 Activation energy

Activation energy, E, kJ/mol
for cardboard
Constituent 1st stage 2nd stage 3rd stage
Cardboard 128.15 133.65 83.74
Fig. 3 Cardboard char (10 °C/min)
The elemental analysis of residues of lignocellulosic waste material, i.e., card-

board is performed with the help of SEM using Energy Dispersive Spectroscopy
(EDS) technique. EDS is a scientific approach and it employs x-rays that are trans-
mitted from the sample when blasted by the electron beam for identifying the
elemental composition of the sample. The elemental analysis of residue with the
help of scanning electron microscope at 10 °C/min of heating rate has been carried
out. The importance of this study is to explore products and elements presence in the
residues of cardboard [13]. Figures 3 and 4 indicate elemental analysis of cardboard
ash obtained from TGA at 10 °C/min of heating rate.
5 Conclusion
For cardboard material, decomposition occurs in three stages. Stage I, II, and III repre-
sents decomposition of hemicellulose, cellulose, and lignin, respectively. The esti-
mation of decomposition rate is possible through the assurance of kinetic parameters
such as activation energy.
The activation energy for cardboard waste at the heating rate of 5 °C/min to 15
°C.
/min is obtained as 128.15 kJ/mol in first stage, 133.65 kJ/mol in second stage
and 83.74 kJ/mol in third stage, respectively.
The significance of elemental analysis is to know the possible elements existing
in the residue and how they will affect the environment from pollution point of view
in later stage.
Fig. 4 Elemental spectrum
6 Future Scope of Work
The present research work can be extended for development of such reactor at
macro-level of cardboard or any MSW constituents for various heating rate makes
possible. It also yields products obtained by such reactor which establishes MSW as
an alternative better fuel.
References
1. Garcia AN, Marcilla A, Font R (1995) Thermogravimetric kinetic study of the pyrolysis of
municipal solid waste. Thermochim Acta 254:277–304
2. Font R, Marcilla A, Garcia AN, Caballero JA, Conesa JA (1995) Comparison between the
pyrolysis products obtained from different organic wastes at high temperatures. J Anal Appl
Pyrol 32:41–49
3. Wu C-H, Chang C-Y, Lin J-P, Hwang J- Y (1997) Pyrolysis kinetics of paper mixture in
municipal solid waste. Fuel 76:115l–1157
4. Park IW, Sea Cheon O, Lee HP, Taik Kim H, Yoo KO (2000) A kinetic analysis of thermal
degradation of polymers using a dynamic method. Polym Degrad Stabil 67:535–540
5. Sorum L, Gronli MG, Hustad JE (2001) Pyrolysis characteristics and kinetics of municipal
solid waste. Fuel 80:1217–1227
6. David C, Salvador S, Dirion JLM (2007) Determination of a reaction scheme for cardboard
thermal degradation using thermal gravimetric analysis. J Anal Appl Pyrolysis 67:307–323
7. Molto J,Font R, Conesa A, Martıń-Gulloń I (2005) Thermogravimetric analysis during the

decomposition of cotton fabrics in an inert and air environment. J Anal Appl Pyrolysis (Article
in Press)
8. Miranda R, Sosa_Blanco C,Bustos-Martı ńez D, Vasile C (2007) Pyrolysis of textile wastes
kinetics and yields. J Anal Appl Pyrolysis 80:489–495
9. Buah WK, Cunliffe AM, Williams PT (2007) Characterization of products from the pyrolysis
of MSW. Process Saf Environ Prot 85:450–457
10. Yourlmaz S (2006) Investigation of emissions and combustion kinetics of waste wood samples
with thermal and spectral methods. PG Dissertation, Middle East Technical University, Beirut
11. Liu Z (1998) Handbook of thermal analysis. John willy & Sons, pp 47–48
12. Flynn JH, Wall LA (1966) Polym Sci Part B, Polymer Letters 4:323
13. Miskolczi N, Angyal A, Bartha L, Valkai I (2009) Fuels by pyrolysis of waste plastics from
agricultural and packaging sectors in a pilot scale reactor. Fuel Process Technol 90:1032–1040
Chapter 5
Wavelength Optimization in Gigabit
Passive Optical Network by Proposed
Quad Play Architecture
1 Introduction
Unruly modification in the enterprise resident area network is the prime cause
to fulfill the requirements of swift progress of high bandwidth requirements and
also changing the nature of enterprise solutions. To maintain a significant solution
and to reduce the capEX and opEX, fulfilling the high bandwidth demand, the
utmost solution is gigabit passive optical network (GPON) technology [1–2]. ITU-T
G.984.x approvals offer GPON system prototypical and does not need any electrical
power at the transitional nodes between the aggregation and last-mile device [3].
A big number of researches had been worked on network structures, propagation
policies, power consumption budget, bandwidth distribution, and stability of GPON
technology [4–6]. The efficiency of service for GPON triple play in IP network and
also passive optical network and bandwidth distribution policies are conversed in
[7–8]. Ricciardi et al. have exposed an investigation and contrast between EP2P
(Ethernet point to point) and GPON solution structure [9].
SDH (E1) traffic with the present triple play explanation is the demand of today’s
technological solution. It is much more important to focus on some parameters like
output signal range, propagation wavelength, power budget, and sensitivity threshold
for convergence solution. In this paper, wavelength is optimized with a proposed
quad play architecture by calculating different parameters like path loss, the last end
receiving sensitivity, etc.
The rest of this paper is ordered as follows. The literature review is presented in
Sect. 2. GPON triple play general structure is stated in Sect. 3. In Sect. 4, proposed
Md. H. Ali (B) · M. H. Ali

Jahangirnagar University, Dhaka, Bangladesh
e-mail: hayder.iict@gmail.com
M. H. Ali
e-mail: hanif_ju03@juniv.edu

https://doi.org/10.1007/978-981-15-8820-4_5
54 Md. H. Ali and M. H. Ali
quad play architecture is briefly defined. The simulation scenario is in Sect. 5 and
performance analysis between GPON triple play and proposed quad play is stated in
Sect. 6. Finally, Sect. 7 draws a conclusion to this paper.
2 Literature Review
Over the years especially in the last decade, a lot of research study have been passed
out to find out the GPON network suitable structure and different type of FTTx
model. Some of them are discussed below.
Koonen et al. [10] stated fiber-optic technologies and it enables the solution with
minimum cost of broadband solution services to the last mile users in multiple-access
network architectures [10]. They also designated that the dynamic network recon-
figuration can improve system staging. Derek Nesset et al. [11] have investigated
GPON structure with a 1300 nm semiconductor optical amplifier and it developed
for long spread GPON services [11]. The high gain of −29 dB has empowered
a profitable GPON system to function over 60 km and with 128-way split. Claudio
Rodrigues et al. [12] have studied and associated the present GPON fiber to the home
(FTTH) type service solution, from the point of view of enterprise operator, they also
stated that future next-generation passive optical networks (PONs) (XGPONs) and
wavelength division multiplexing PONs (WDM-PONs). They also consider some
facts like considering into account standardization, wavelength planning, optical
line terminal (OLT) as well as optical network terminal (ONT) equipment, and
transmission convergence layer [12].
Marcelo Alves Guimaraes et al. [13] have projected a new system for E1 transmis-
sion and it allows the fair bandwidth policy, it is performed without circuit emulation
techniques by fragmenting E1 signals. Andrej Chu et al. [14] presented an expla-
nation by using ant cluster optimization to strategy GPON-FTTH networks with
combining equipment [14]. Due to the reflection of many strategy issues such as
the number, types, positions of network elements, the optical network development
progression frequently parades numerous trials from the optimization point of view.
Irfan imitation an estimation of 2.5 Gbps bi-directional GPON based FTTH link
using innovative modulation setups [15]. In this research, the presentation of the
future scheme was calculated when a single wavelength and two wavelengths were
used for triple-play services with different modulation systems.
3 Present Structure of GPON Triple Play
In the present GPON triple play structure, the signals enter intoOLT in two separate
ways. Firstly, EDFA receives a video signal and pass through. Secondly, data and
voice signals enter via a layer-2 switch (ISP connection). These two signals modu-
lated into OLT and passed through fiber, optical splitter up to ONT. Finally, end-users
receive signals (data, video, and voice) from ONT’s ethernet port (Fig. 1).
5 Wavelength Optimization in Gigabit Passive Optical Network … 55
EDFA Voice
Data
ISP ONT
LDP Video
Splitter
SW
OLT
BDB ONT
Optical Fiber
UTP Cable
Coaxial Cable
Fig. 1 Present triple play architecture
4 Projected Quad Play Architecture
The projected quad play design is not similar as present triple play design. All signals
(video and ISP’s data and voice) will enter into a MUX combinedly. These signals
will be modulated into OLT. From the OLT, the modulated signal will propagate
toward the ONT. The end-users will receive the signals from ONT’s ethernet port as
present triple play.
The main difference between the present triple play design and projected quad
play design is, the modulation done into OLT in the present design and OLT becomes
busier and OLT’s CPU uses become high, On the other hand, modulation is done
into MUX and that’s why OLT does not disburse time for the modulation (Fig. 2).
5 GPON Quad Play Simulation Environment
For better results, the simulation experimented only for single-user connectivity. For
this goal, the packet simulation software OptSIM has been used. OptSIM proposes a
hierarchical structure of pattern for creating new simulation scenarios. Defined tree
levels:
• Network model: It’s the primary stage of design. It’s the maximum intellectual and
general level. The goals will be to describe network topology, to express network
nodes, and to define the statement between each node.
• Node model: The second level, objectives are to express the functionalities for
each node that is used in the network topology. For all nodes, we built a scheme
for scheming inner purposes with the component offered by OptSIM.
Voice
Data
Video
Video ONT
Signal
LDP SDH
(MUX)
OLT
Splier
MUX
BDB
ISP ONT
Opcal Fiber
UTP Cable
Fig. 2 Proposed quad play architecture
• Process model: The final level. In this level, a graph is defined for elements used
in the second level. The graphs stipulate the jobs performed by the module with
the info which will be processed.
5.1 Network Construction
To generate the simulation scenario, it is used the tree levels proposed by OptSim.
We have agreed to use FTTH topology (no FTTC or others) because the optical fiber
reaches straight at the user’s home and it is the resolution that offers more bandwidth
capacity.
It is considered only this type of control packets in our simulation and ignored
others feature related to delays or protocols since the objectives of our analysis
need only results about wavelength optimization due to intermediate entree control
measures.
5.2 ONU/ONT Design
User’s ONT contains splitter, and data earpiece, and video earpieces. Data earpiece
is constructed with optical filter, PIN/TIA receiver, and BER Tester. The video signal
receiver contains optical filter, PIN/TIA receiver, and electrical filters (Fig. 3).
Spectrum Plot-1 Spectrum Plot-2
Electrical
Multi plot-1
Filter-1 Video Optical RX
Optical Filter-1 Receiver-1 Electrical

Multi plot-2
Filter-2
BER Tester
Optical Filter-2 Receiver-2
Data/Voice RX
Optical Splitter Multi plot-3

Spectrum Plot-3
BER Tester
Optical Filter-2 Receiver-2
SDH Traffic
Receiver
Multi plot-3
Spectrum Plot-3
Fig. 3 Block diagram of ONU/ONT
5.3 OLT Design
First, we describe in detail the configurations for Central Office OLT and single end-
user ONT and then, will generalize the treatment to all 16 uses. In this experiment, we
consider the downstream configuration of GPON with bitrate 1.25 Gb/s and support
for quad play. The quad play provision is appreciated as a blend of SDH traffic, data,
voice, and video indications. The high-speed Internet element is denoted by a data
link with 1.25 Gb/s downstream bandwidth. The voice element can be characterized
as VOIP service (voice over IP, packet-switched protocol) and can be collected with
data component in physical layer simulations. Finally, the video element can be
denoted as an RF video signal (traditional CATV) or as an IPTV signal that also can
be mutual with data. In this research, it is considered the former case with an RF
video link. To improve the wavelength in PON, the propagation over the optical fiber
pathway services the CWDM technique with data/voice component transmitted at
wavelengths in the variety of 14,801,500 nm, and video within the 1550–1560 nm
range [16] (Fig. 4).
Video Optical TX 1.55 um Signal Plot-1 Signal Plot-2
Spectrum Plot-3
Sine Wave Spectrum Plot-1 Spectrum Plot-2
Generator
Signal Plot-3
Summer DML-1 Pre-Amplifire-1
Sine Wave
Generator
RF Video Signal Optical MUX

(55-1000 MHz)
Electrical
BER Tester DML-2 Pre-Amplifire-2
Generator
Data / VOIP TX Signal Plot-4 Spectrum Plot-4

1.25 Gbps
Fig. 4 Block diagram of OLT
The Central Office Optical Line Terminal block (Transmitter block) involves
Data/VOIP and Video elements. The Data/VOIP spreader demonstrated with pseudo-
random data originator (PRBS), NRZ modulator driver, direct-modulated laser, and
admirer amplifier. The video section demonstrated as RF SCM (subcarrier multi-
plexed) link with only two tones (channels) for simplicity. The two channels we used
are from standard NTSC analog CATV frequency plan—channel 2 and channel
78 at frequencies 55.25 MHz and 547.25 MHz, respectively. RF video source
contains two electrical signal producers, summer, direct-modulated laser, and pre-
amplifier. Following, data/voice and video signals are multiplexed at multiplexer and
propagated into 20 km fiber span.
6 Result Analysis
Applying the loss calculation equation (in Sect. 5), the below table is founded.
It shows the last end (end-user) devices (ONT/ONU) signal receiving sensitivity.
Minimum receiving sensitivity is the desired one (Table 1).
Using 1:32 Splitter with 20 km distance, it is observed that the minimum sensitivity
is −34.35 dB at 1550 nm wavelength and the highest sensitivity is −36.66 dB at
1310 nm wavelength (Fig. 5).
Table 1 1:32 Splitter with 20KM distance
Wave Attenuation Attenuation Mux Add Drop Demux Filter ODF LDP Splitter BDB ONT Totale Remarks
length (dB/Km) for 10m loss Loss Loss Loss Loss (IN+OUT) (IN+OUT) Loss (IN+OUT) Loss
(dB) (dB) Loss Loss Loss
1310 0.36 7.2 0.5 1 1 0.5 8.5 0.3+0.3=0.6 0.3+0.3=0.6 17.3 0.3+0.3=0.6 0.3 36.66
1330 0.335 6.7 0.5 1 1 0.5 8.5 0.3+0.3=0.6 0.3+0.3=0.6 17.3 0.3+0.3=0.6 0.3 36.135
1350 0.322 6.44 0.5 1 1 0.5 8.5 0.3+0.3=0.6 0.3+0.3=0.6 17.3 0.3+0.3=0.6 0.3 35.862
1370 0.311 6.22 0.5 1 1 0.5 8.5 0.3+0.3=0.6 0.3+0.3=0.6 17.3 0.3+0.3=0.6 0.3 35.631
1390 0.333 6.66 0.5 1 1 0.5 8.5 0.3+0.3=0.6 0.3+0.3=0.6 17.3 0.3+0.3=0.6 0.3 36.093
1410 0.291 5.82 0.5 1 1 0.5 8.5 0.3+0.3=0.6 0.3+0.3=0.6 17.3 0.3+0.3=0.6 0.3 35.211
1430 0.281 5.62 0.5 1 1 0.5 8.5 0.3+0.3=0.6 0.3+0.3=0.6 17.3 0.3+0.3=0.6 0.3 35.001
1450 0.272 5.44 0.5 1 1 0.5 8.5 0.3+0.3=0.6 0.3+0.3=0.6 17.3 0.3+0.3=0.6 0.3 34.812
1470 0.266 5.32 0.5 1 1 0.5 8.5 0.3+0.3=0.6 0.3+0.3=0.6 17.3 0.3+0.3=0.6 0.3 34.686
1490 0.26 5.2 0.5 1 1 0.5 8.5 0.3+0.3=0.6 0.3+0.3=0.6 17.3 0.3+0.3=0.6 0.3 34.56
1510 0.254 5.08 0.5 1 1 0.5 8.5 0.3+0.3=0.6 0.3+0.3=0.6 17.3 0.3+0.3=0.6 0.3 34.434
5 Wavelength Optimization in Gigabit Passive Optical Network …
1530 0.252 5.04 0.5 1 1 0.5 8.5 0.3+0.3=0.6 0.3+0.3=0.6 17.3 0.3+0.3=0.6 0.3 34.392
1550 0.25 5 0.5 1 1 0.5 8.5 0.3+0.3=0.6 0.3+0.3=0.6 17.3 0.3+0.3=0.6 0.3 34.35 Best
1570 0.25 5 0.5 1 1 0.5 8.5 0.3+0.3=0.6 0.3+0.3=0.6 17.3 0.3+0.3=0.6 0.3 34.351
1590 0.257 5.14 0.5 1 1 0.5 8.5 0.3+0.3=0.6 0.3+0.3=0.6 17.3 0.3+0.3=0.6 0.3 34.497
1610 0.266 5.32 0.5 1 1 0.5 8.5 0.3+0.3=0.6 0.3+0.3=0.6 17.3 0.3+0.3=0.6 0.3 34.686
59
Fig. 5 Receiving sensitivity at different splitting ratio with various distance
Analyzing the sensitivity calculation table and graph, it is instituted that 1550 nm
wavelength has sensitivity with minimum dB loss. That is, end-user will get better
performance. Could use 1550 nm wavelength for GPON quad play instead of using
1310–1610 nm.
In general triple play services, the eye diagram could not measure, but in proposed
quad play, can easily measure the eye diagram from ONT/ONU’s ethernet port.
Figure 6 shows that almost noise-free traffic could be captured from quad play
architecture.
In Fig. 7, for data connectivity, the blue stripe displays the general triple play
sensitivity and the green line displays the projected quad play sensitivity. General
triple play has approximately −30 dBm power loss while projected quad play has
approximately −25 dBm.
In Fig. 8, in the state of video connectivity, the blue line demonstrates the general
triple play sensitivity and the green line displays the projected quad play sensitivity.
Fig. 6 Eye diagram of SDH

traffic (QP)
Fig. 7 Data output signal

for quad play versus triple
play for 1550 nm
Fig. 8 Video output signal

for quad play versus triple
play for 1550 nm
Fig. 9 Total transmitted bit versus number of error
The general triple play has approximately −35 dBm power loss while the proposed
quad play has approximately −32 dBm.
From the above Fig. 9, it is found that for every 1 megabit transmission there is
64-bit error in projected quad play while 76-bit error in a general triple play. If the
quantity of transmitted bit increased, then the number of errors will also increase in
triple play, but in the projected quad play the error bit will not increase in the same
ratio compare to triple play.
Form the above Table 2, it can be summarized that if we use 1550 nm wave-
length for GPON quad play services, we gain approximately −5 dB for data and
voice, approximately −3 dB for video and these are the best gain compare to other
wavelengths at GPON triple play service area.
7 Conclusion
The main attention was to learn the best wavelength for GPON quad play solution
with high performance. Nowadays, the usual triple play facility solution is operating
by a varied range of wavelengths. Here, 1310 nm to 1550 nm wavelength is used
for triple play connectivity. Sometimes, it makes some intrusion within consumers.
It was keen to find out a single wavelength as an alternative of numerous wave-
lengths. Analyzing all possible combinations, it is initiated that for quad play archi-
tecture, 1550 nm wavelength gives a better eye diagram for SDH traffic and better
performance in data, voice, and video service.
Table 2 Comparison of different wavelength with dB gain in Quad Play

SN Wavelength (nm) Input Output (dBm) Gain Remarks
(dBm) (dBm)
Voice Data Video V= Triple Quad (TP-QP)
Video Play Play
(TP) (QP)
1 1310 1310 1310 0, −10 −30, − −27, − 3, 0 (V)
(V) 26 (V) 26 (V)
2 1490 1490 1490 0, −10 −30, − −30, − 0, 2 (V)
(V) 32 (V) 30 (V)
3 1550 1550 1550 0, −10 −30, − −25, − 5, 3 (V) Best
(V) 32 (V) 29 (V)
4 1490 1490 1310 0, −10 −34, − −32, − 2, 6 (V)
(V) 29 (V) 35 (V)
5 1550 1550 1310 0, −10 −35, − −30, − 5, 2 (V)
(V) 29 (V) 27 (V)
6 1490 1490 1550 0, −10 −35, − −38, − 3, 7 (V)
(V) 34 (V) 27 (V)
References
1. Cale I, Salihovic A, Ivekovic M (2007) Gigabit passive optical network—GPON. In:

Proceedings of the 29th international conference on information technology interfaces, pp
25–28
2. Leo M, Trotta M (2011) Performance evaluation of WDM-PON RSOA based solutions in
NGAN scenario. In: Proceedings of the 50th FITCE congress (The Forum for European ICT
and Media Professionals), pp 1–4
3. ITU-T Recommendation G.984.1 (2003), 2 (2003), 3 (2004), 5 (2006) and 6 (2007).
4. Frnda J, Voznak M, Fazio P, Rozhon J (2015) Network performance QoS estimation. In:
Proceedings of the 38th international conference on telecommunications and signal processing
(TSP), pp 1–6
5. Rokkas T (2015) Techno economic analysis of PON architectures for FTTH deployments. In:
Proceedings of the conference of telecommunication, media and internet techno-economics
(CTTE), pp 1–5
6. Lee SS, Li W, Wu MS (2016) Design and implementation of a GPON-based virtual open
flow-enabled SDN switch. J Lightwave Technol 34(10):1–8
7. Lee J, Hwang I, Nikoukar A, Liem T (2013) Comprehensive performance assessment of
bipartition upstream bandwidth assignment schemes in GPON. J Optic Commun Netw
5(11):1285–1295
8. Milanovic S (2014) Case study for a GPON deployment in the enterprise environment. J Netw
9(1):42–47
9. Ricciardi S, Santos-Boada G, Careglio D, Domingo-Pascual J (2012) GPON and EP2P: a
techno-economic study. In: Proceedings of the 17th European conference on networks and
optical communications (NOC), pp 1–6
10. Koonen AMJ (2006) Technologies and applications of FTTx. Published at 19th annual meeting
of the IEEE, lasers and electro-optics society, LEOS 2006
11. Nesset D, Appathurai S, Davey R (2008) Extended reach GPON using high gain semiconductor
optical amplifiers. In: Proceedings of the conference on optical fiber communication/national
fiber optic engineers (OFC/NFOEC), pp 8–16
12. Rodrigues C, Gamelas A, Carvalho F, Cartaxo A (2011) Evolution of FTTH networks based
on radio-over-fibre. In: Proceedings of the 13th international conference on transparent optical
networks, pp 7–12
13. Guimarães MA, Rocha LD (2011) Fractional E1 transport in gigabit passive optical network.
In: Proceedings of the SBMO/IEEE MTT-S international microwave and optoelectronics
conference (IMOC 2011), pp 44–50
14. Chu A, Poon KF, Ouali A (2013) Using ant colony optimization to design GPON-FTTH
networks with aggregating equipment. In: Proceedings of the IEEE symposium on computa-
tional intelligence for communication systems and networks (CIComms), pp 1–6
15. Róka R (2015) Analysis of possible exploitation for long reach passive optical networks. In:
Proceedings of the 4th international conference on simulation and modeling methodologies,
technologies and applications (SIMULTECH), pp 123–130
16. OptSIM user manual, Available online: https://www.lmd.polytechnique.fr
Chapter 6
A Priority-Based Deficit Weighted Round
Robin Queuing for Dynamic Bandwidth
Allocation Algorithm in Gigabit Passive
Optical Network
1 Introduction
Accessing Internet is going to be a fundamental right like other basic human rights.
It is not a matter that how often a user uses or surf Internet, waiting for a web page
to load is certainly an irritation. To overwhelm this condition, Internet surfing speed
has enlarged meaningfully in the earlier era to retain step with the petition of end-
users, innovative amenities, and bandwidth-hungry claims. These anxieties embrace
as hypermedia content-based e-commerce, video on claim, high classification TV,
IPTV, online gaming, social media, etc. The communication protocols for hyper-
media circulation have conventional a great transaction of attention in the earlier
few ages. Since multimedia traffic must provision numerous types of traffic simulta-
neously, it is crucial to process data according to its characteristics. Thus, protocol
originators have to grasp the features of traffic and select a processing method suit-
able for the performance requirements. For instance, actual acoustic traffic in a voice
service requires rapid transmission, but the loss of a small amount of audio infor-
mation is tolerable. On the opposite site, the transfer of a text file should guar-
antee 100% reliable transfer; real-time delivery is not of primary importance in this
case. Real-time video service, such as video on demand (VOD), requires not only
rapid transfer but also high reliability. When a piece of video information is lost, its
quality of service (QoS) is degraded. Therefore, multimedia communication proto-
cols should be premeditated to afford the performance requirements of a wide range
of multimedia services [1–7].
Md. H. Ali (B) · M. H. Ali

Jahangirnagar University, Dhaka, Bangladesh
e-mail: hayder.iict@gmail.com
M. H. Ali
e-mail: hanif_ju03@juniv.edu

https://doi.org/10.1007/978-981-15-8820-4_6
The other parts of this paper are organized as follows. Proposed DBA poling
mechanism for PMDWRR is introduced in Sect. 2. In Sect. 3, Pseudo code for
PMDWRR is briefly described. Simulation scenario is in Sect. 4 and performance
analysis between WRR, DWRR, and PMDWRR and stated in Sect. 5. Finally, Sect. 6
draws a conclusion to this paper.
2 Proposed DBA Poling Mechanism Design for Modified

Deficit Weighted Round Robin
1. Start
2. ONUs/ONTs calculate the amount of received data
3. ONUs/ONTs predict the size of data in waiting time
4. ONUs/ONTs calculate the traffic for T-CONTs ( T-CONT1 to T-CONT4)
5. ONUs/ONTs enhance the prediction of accuracy by calculating arriving in
waiting time (PT i,n ) for n-th Cycle.
6. OLT calculates the arithmetic mean for the data received in waiting time.
7. Repeat step 2 to 6 for n-th cycle.
8. Then
9. Calculate the difference for the size of data received in waiting time at
ONTi /ONUi in the n-the cycle.
10. Calculate the size of traffic prediction arriving in waiting time at ONTi /ONUi
in the n-th cycle by adding a weighting factor.
11. Increase the average order to ensure the prediction accuracy.
12. OLT grant frame size or allocate the frame size for priority traffic.
13. ONTs/ONUs send data according to OLT’s priority grant frame.
14. End
To execute the above algorithm for GPON networks, it is necessary to maintain

a queuing system. DWRR algorithm is suitable for this. As it provides protection
among different flows, it overcomes the limitations of strict PQ (Fig. 1).
3 Pseudo Code for Priority-Based Deficit Weighted Round

Robin
The pseudo code in this section does not describe the procedure of any exact DWRR
execution. Although each execution will vary from this model, studying the samples
and outlining the pseudo code will make it calmer to recognize the explicit enterprise
results that are compulsory to make for executions.
The array flexible Deficit_Counter is reset to zero. In this sample, the queues are
numbered 1 to n, where n is the extreme number of lines on the output port:
6 A Priority-Based Deficit Weighted Round Robin … 67
OLT
ONUi/ONTi t1 t2Waing Time t3

REPORT Message
An Cycle GATE Message
Fig. 1 DBA polling mechanism diagram for PMDWRR
FOR i = 1 to n /* Visit each queue index */

Deficit_Counter[i] = 0 /* Initialize Deficit_Counter[i] to 0 */
ENDFOR
i = the index of the queue that will grip the new package
IF (ExistsInActiveList(i) = FALSE) THEN /*IF i not in
ActiveList */ InsertActiveList(i) /* Add i to the end
of ActiveList */ Deficit_Counter[i] = 0 /*Reset
queue Deficit_Counter[i] to 0*/
ENDFOR
Enqueue packet to Queue[i] /* Place packet at finish of queue i */
END Enqueue
// Removes the element with the highest priority form the list
void pop (Node** head)
{
Node* temp = *head;
(*head) = (*head)->next;
free(temp);
}
// Function to push according to priority

void push (Node** head, int d, int p)
{
Node* start = (*head);
Whenever an index is at the head of the ActiveList, the function Dequeue ()
transmits up to Deficit_Counter[i] + Quantum[i] worth of bytes from queue.
Dequeue ()
While (TRUE) DO
IF (ActiveList is NotEmpty) THEN
i = the index at the head of the ActiveList
Deficit_Counter[i] = Deficit_Counter[i] +
Quantum[i]
WHILE (Deficit_Counter[i] > 0 AND NOT
Empty (Queue[i])) DO
PacketSize = Size (Head (Queue[i]))
IF (PacketSize <= Deficit_Counter[i]) THEN
Transmit packet at head of Queue[i]
Deficit_Counter[i] = Deficit_Counter[i] –
PacketSize
/* To find out the largest traffic volume */
/*Declare largest as integer*/
Set largest to 0
FOR EACH value in A DO
IF A[n] is greater than largest THEN
largest A[n]
ENDIF
END FOR
Dequeue the largest first
ELSE
Break /*exit this while loop*/
ENDIF
ENDWHILE
IF (Empty (Queue[i])) THEN Deficit_Counter[i] = 0
RemoveFromActiveList(i)
ELSE
InsertActiveList(i)
ENDIF
ENDIF
ENDWHILE
END Dequeue
4 Simulation Design
The simulation is done by using OptSIM (RSoft System Suit, version-2016.06) Simu-
lation software. The simulation scenario is like bellow Fig. 2. It is composed of one
OLT (Optical Line Terminal), one 1:32 Splitter and then thirty-two ONUs/ONTs.
Each ONU/ONT has four signal sources producing T-CONT1 to T-CONT4 services
respectively. Before each ONT, there is Bandwidth Analyzer to measure the
consumed bandwidth of individual ONU/ONT. The OLT’s up and down link rates
are 1.25 and 2.48 Gbps. Each ONU/ONT is 20 km far from the OLT.
Fig. 2 Simulation scenario BA
in OptSIM simulator
BA ONT-1
BA
ONT-2
ONT-3
Splier
OLT
BA ONT-32
5 Result Analysis
The data has captured from OptSIM simulator and analyzed by MATLAB. Delay
measurement and usage bandwidth are calculated by MATLAB coding.
Figure 3 stated that number of end-users is more in PMDWRR. Considering
2GBPS bandwidth allocation, DWRR is allowing 17 users, MDWRR is allowing 20
users, while PMDWRR is allowing 23 users. Bandwidth is shared with maximum
users which satisfy the properties of GPON system.
Figure 4 shown that delay per user is minimum in PMDWRR. Considering 25
users in the system, DWRR shows 3 ms delay, MDWRR shows 2.4 ms delay while
Fig. 3 Bandwidth allocation

versus number of users
Fig. 4 Delay versus number

of users
PMDWRR shows only 1.7 ms/user. The per user delay is less in PMDWRR and it
increases the system performance.
6 Conclusion
In this paper, have partially modified to set priority into the existing MDWRR. In
PMDWRR, ONU/ONT predict the data size and send allocation request to OLT,
OLT increases the prediction accuracy. It could calculate the size of the queue and
store the data size. By calculating the queue size, it can predict the priority traffic. It
transmits the priority of the traffic and this way it could minimize the transmission
delay and it satisfies the full service of QoS requirements of GPON system.
References
1. Tae Il J, Jae Ho J, Sung Jo K (1997) An efficient scheduling mechanism using multiple thresholds
for multimedia traffic in ATM switching nodes. In: Proceedings of the 22nd IEEE conference
on local computer networks (LCN’97)
2. Michael P Mc, Martin M, Martin R (2004) Ethernet PONs: a survey of dynamic bandwidth
allocation (DBA) algorithms. IEEE Commun Mag 42(8), 1–15
3. Taeck-Geun K, Sook-Hyang L, June-Kyung R (1998) Scheduling algorithm for real-time
burst traffic using dynamic weighted round robin. In: Proceedings of the IEEE international
symposium on circuits and systems (Cat. No.98CH36187), pp 506–509
4. Patel Z, Dalal U (2014) Design and implementation of low latency weighted round robin
(LLWRR) scheduling for high speed networks. Int J Wirel Mob Netw (IJWMN) 6(4):59–71
5. Ji-Young K, Ji-Seung N, Doo-Hyun K (2002) A modified dynamic weighted round robin cell
scheduling algorithm. ETRI J 24(5):360–372
6. Ouni R, Bhar J, Torki K (2013) A new scheduling protocol design based on deficit weighted
round robin for QoS support in ip networks. J Circ Syst Comput 22(3):1–21
7. Lenzini L, Mingozzi E, Stea G (2006) Bandwidth and latency analysis of modified deficit round
robin scheduling algorithms. In: Proceedings of the 1st international conference on performance
evaluation methodologies and tools (VALUETOOLS 2006)
Chapter 7
Protection of Six-Phase Transmission
Line Using Demeyer Wavelet Transform
Gaurav Kapoor
1 Introduction
An increase in the inevitability of electrical power has been perceived by the people
of the modern generation. The electrical power transfer potentiality of the currently
operating power transmission systems ought to be augmented in order to assist the
significant increase in the necessity of electrical energy.
Thus, accurate recognition of the faults in the SPTL turns out to be very decisive
for mitigating the loss of gain and providing fast renovates.
Numerous newly reported research works addressed the issue of FD and FC in
TL’s. The discussion of few researches is shown in brief here in this section. In [1],
FSM-based technique has been used for TL protection. In [2, 4, 7], WT and MM have
been applied for the protection of SPTL. VMD has been employed for disturbance
recognition in power system [3]. POVMD and WPNRVFLN have been used for
FD and FPR in SCDCTL [5]. DM and MM have been used for high impedance
fault recognition in TL [6]. Alienation-based technique has been proposed for TTTL
protection [8]. In [9], the theory of TW has been reported and the combination of
VMD and TEO has been applied for HVDC TL protection. VMD and HT have been
used for micro-grid protection [10].
In this work, the demeyer wavelet transform (DMWT) has been used for the
protection of six-phase transmission line (SPTL). No such type of work has been
reported yet to the best of the knowledge of author. The results exemplify that the
DMWT efficiently recognizes and categorizes the faults and the consistency of the
DMWT is not affected by variation in the fault factors of SPTL.
G. Kapoor (B)
Modi Institute of Technology, Kota, India
e-mail: gaurav.kapoor019@gmail.com

https://doi.org/10.1007/978-981-15-8820-4_7
74 G. Kapoor
Load-1
Six Phase
138 kV source Transmission 138 kV source
Relay Line
Load-2
Bus-1 Fault Bus-2
Fig. 1 The schematic of SPTL
This paper is structured as follows: The specifications of SPTL are presented in

Sect. 2. The flow-diagram for DMWT is presented in Sect. 3. Section 4 is dedicated
to the discussion of results. Section 5 concludes the paper.
2 The Specifications of SPTL
Figure 1 shows the schematic of SPTL. The schematic consists of 138 kV, 60 Hz
SPTL of 68 km, connected to a 138 kV voltage source at one end and loads at the
other end. The SPTL is divided into two parts of length 34 km each. The CT’s and
relay are connected at bus-1 to protect the full length of SPTL.
3 The Flow Chart of DMWT
Figure 2 illustrates the process for the DMWT. The steps are shown beneath.
Step 1 Simulate the SPTL for creating faults and produce the post-fault currents.
Step 2 Use DMWT to examine the post-fault currents for characteristics retrieval
and determine the range of DMWT output.
Step 3 The phase will be proclaimed as the faulted phase if its DMWT output has
a larger amplitude as compared to the healthy phase in a faulty situation.
4 Performance Assessment
The simulations have been performed for the near-in relay faults, far-end relay faults,
converting faults, inter-circuit faults, and multi-position faults with the objective of
verifying the feasibility of the DMWT. In the separate subcategories, the outcomes
of the work are investigated.
7 Protection of Six-Phase Transmission Line … 75
Record six
phase current
Signal processing using DMWT
Evaluate DMWT
coefficients
Is |DMWT coefficient| faulted phase No

> No fault
|DMWT coefficient| healthy phase
Yes
Fault detection
and
Faulty phase recognition
Fig. 2 Flow diagram showing DMWT process
4.1 The Efficacy of DMWT for Healthy Condition
Figure 3 shows the SPTL currents and voltages for no-fault. The DMWT outputs
for no-fault are shown in Fig. 4. Table 1 reports the results of DMWT for healthy
situation.
Fig. 3 Six-phase current and voltage waveforms for no-fault

76 G. Kapoor
a b c
d e f
Fig. 4 DMWT outputs of six-phase currents for no-fault
Table 1 Results of DMWT for no-fault

DMWT output
Phase-A Phase-B Phase-C Phase-D Phase-E Phase-F
175.9988 152.7358 155.3587 154.5936 133.9446 188.1193
4.2 The Efficacy of DMWT for Near-in Relay Faults
The efficiency of the DMWT is investigated for the near-in relay faults on the SPTL.
Figure 5 exemplifies the SPTL currents for ABDFG near-in relay fault at 5 km at
0.0535 s among RF = 2 and RG = 4 . Figure 6 illustrates the DMWT output for
ABDFG fault. The fault factors for the fault cases are: T = 0.0535 s, RF = 2 , and
Fig. 5 ABDFG near-in relay fault at 5 km at 0.0535 s among RF = 2 and RG = 4

a b c
d e f
Fig. 6 DMWT output for ABDFG near-in relay fault at 5 km at 0.0535 s
Table 2 Results of DMWT for near-in relay faults

Fault type DMWT output
ABDFG 975.5428 1.5645 × 82.1565 905.7343 62.6620 1.4003 ×
(5 km) 103 103
ABCG 1.1302 × 1.0896 × 1.2082 × 86.5100 44.6221 50.1690
(6 km) 103 103 103
BCEFG 208.6318 1.1057 × 1.0163 × 151.0041 1.3309 × 1.0599 ×
(7 km) 103 103 103 103
ABCDEG 1.4257 × 1.1870 × 1.1547 × 1.1865 × 993.3149 208.7100
(8 km) 103 103 103 103
DEFG 53.2673 46.4882 60.4460 1.9163 × 1.8024 × 1.1828 ×
(9 km) 103 103 103
RG = 4 . Table 2 details the results of the DMWT for near-in relay faults. It is seen
from Table 2 that the DMWT precisely detects the near-in relay faults.
4.3 The Efficacy of DMWT for Far-End Relay Faults
The DMWT has been explored for the far-end relay faults. Figure 7 illustrates the
SPTL currents for ABCEFG far-end relay fault at 63 km at 0.1 s among RF =
1.15 and RG = 2.15 . Figure 8 shows the DMWT output for ABCEFG fault.
Table 3 reports the results for far-end relay faults. It is inspected from Table 3 that
the effectiveness of the DMWT remains unaffected for the far-end relay faults.
78 G. Kapoor
Fig. 7 ABCEFG far-end fault at 63 km at 0.1 s among RF = 1.15 and RG = 2.15
a b c
d e f
Fig. 8 DMWT output for ABCEFG far-end relay fault at 63 km at 0.1 s
Table 3 Results of DMWT for far-end relay faults

Fault type DMWT output
ABCEFG 3.8060 × 7.9079 × 4.0745 × 158.1244 3.4767 × 2.5117 ×
(63 km) 103 103 103 103 103
ABCG 4.8194 × 8.5085 × 4.3048 × 165.9761 157.4620 197.3903
(64 km) 103 103 103
DEFG 142.1903 137.1558 124.3455 5.1810 × 2.7625 × 4.8779 ×
(65 km) 103 103 103
ABDEG 5.8668 × 6.2243 × 166.0415 5.8480 × 4.5843 × 153.9397
(66 km) 103 103 103 103
ABCDG 3.0659 × 6.4599 × 3.7578 × 3.1028 × 197.5127 211.9889
(67 km) 103 103 103 103
Fig. 9 SPTL currents for multi-position ABG fault at 42 km and DEFG fault at 26 km
a b c
d e f
Fig. 10 DMWT output for multi-position ABG fault and DEFG fault
4.4 The Efficacy of DMWT for Multi-position Faults
The DMWT is tested for different cases of multi-position faults. Figure 9 depicts the
currents when the SPTL is simulated for the multi-position ABG fault at 42 km and
DEFG fault at 26 km at 0.0725 s among RF = 1.5 and RG = 1.25 . Figure 10
shows the DMWT output for ABG and DEFG multi-position fault. Table 4 presents
the results for different position-position faults.
4.5 The Efficacy of DMWT for Inter-circuit Faults
The DMWT is tested for different cases of inter-circuit faults. Figure 11 depicts the
currents when the SPTL is simulated for the inter-circuit ABCG and FG fault at
30 km at 0.1325 s among RF = 2.75 and RG = 1.85 . Figure 12 depicts the
DMWT output for ABCG and FG inter-circuit fault. Table 5 reports the results for
80 G. Kapoor
Table 4 Results of DMWT for multi-position faults

FT-1 FT-2 DMWT output
(km) (km) Phase-A Phase-B Phase-C Phase-D Phase-E Phase-F
ABG DFFG 4.1866 × 4.4410 × 203.0695 6.5372 × 2.9300 × 6.4875 ×
(42) (26) 103 103 103 103 103
CG DFG 248.1627 247.6852 1.4210 × 2.4488 × 179.0262 2.7339 ×
(20) (48) 103 103 103
DEFG BG 93.2050 2.7384 × 114.9325 4.1065 × 2.7835 × 4.5857 ×
(35) (33) 103 103 103 103
ACG EFG 2.4514 × 111.0308 2.0142 × 132.6308 2.8471 × 4.2729 ×
(44) (24) 103 103 103 103
DG ABCG 4.8020 × 6.5308 × 2.4552 × 2.6415 × 97.7596 163.331
(38) (30) 103 103 103 103
Fig. 11 SPTL currents for ABCG and FG inter-circuit fault at 30 km at 0.1325 s
a b c
d e f
Fig. 12 DMWT output for ABCG and FG inter-circuit fault at 30 km at 0.1325 s

Table 5 Results of DMWT for inter-circuit faults

Fault-1 Fault-2 DMWT output
ABCG FG 5.4287 × 5.0592 × 3.2656 × 159.5320 95.4246 3.3819 ×
103 103 103 103
BG DG 264.7818 4.0431 × 260.5588 3.2120 × 244.3075 245.0540
103 103
ACG DEG 3.7037 × 176.3243 2.6696 × 4.8352 × 3.5412 × 131.1642
103 103 103 103
ABG EFG 5.7039 × 5.8627 × 195.1929 201.4383 2.6028 × 4.5607 ×
103 103 103 103
CG DEFG 271.4223 202.1157 3.9056 × 1.0169 × 8.9809 × 4.5267 ×
103 103 103 103
the inter-circuit faults. It is examined from Table 5 that the DMWT performs well
for the recognition of inter-circuit faults.
4.6 The Efficacy of DMWT for Converting Faults
The DMWT has been investigated for the converting faults. Figure 13 exemplifies
the currents of the SPTL when initially the ABG fault at 35 km at 0.05 s is converted
into the DEFG fault at 35 km at 0.15 s among RF = 2.50 and RG = 1.50 .
Figure 14 exemplifies the DMWT output for the same fault case. The fault factors
preferred for the additional fault cases are: FL = 35 km, RF = 2.50 and RG = 1.50
. Table 6 reports the results for different converting faults.
Fig. 13 SPTL current when ABG fault at 0.05 s is converted into DEFG fault at 0.15 s at 35 km
among RF = 2.50 and RG = 1.50
82 G. Kapoor
a b c
d e f
Fig. 14 DMWT output when ABG fault is converted into DEFG fault
Table 6 Results of DMWT for converting faults

Fault Converted DMWT output
type fault Phase-A Phase-B Phase-C Phase-D Phase-E Phase-F
ABG DEFG 816.6582 1.2847 × 196.2113 7.2091 × 3.1614 × 6.2976 ×
(0.05) (0.15) 103 103 103 103
DG DEFG 159.9643 136.8911 198.3681 5.6853 × 1.7459 × 4.2094 ×
(0.1) (0.2) 103 103 103
ABCG EG 1.5407 × 2.1744 × 1.6289 × 197.2483 1.3585 × 173.1067
(0.07) (0.16) 103 103 103 103
ACG DEG 2.0178 × 171.0564 1.1652 × 3.2852 × 2.0547 × 110.2217
(0.05) (0.2) 103 103 103 103
EFG ABG 5.8620 × 6.6895 × 208.7799 645.9092 1.4708 × 912.3268
(0.1) (0.15) 103 103 103
5 Conclusion
The demeyer wavelet transform (DMWT) is seemed to be very efficient under varied
fault categories for the SPTL. The DMWT output of fault currents of the SPTL
is assessed. The fault factors of the SPTL are varied and it is discovered that the
variation in fault factors do not influence the fidelity of the DMWT. The outcomes
substantiate that the DMWT has the competence to protect the SPTL beside different
fault categories.
References
1. Yadav A, Swetapadma A (2016) A finite-state machine based approach for fault detection
and classification in transmission lines. Elect Power Component Syst (Taylor and Francis)
44(1):43–59
2. Kapoor G (2018) Six phase transmission line boundary protection using wavelet transform.
In: Proceedings of the 8th IEEE India international conference on power electronics (IICPE),
Jaipur, India (2018)
3. Jena MK, Samantaray SR, Panigrahi BK (2017) Variational mode decomposition-based
power system disturbance assessment to enhance WA situational awareness and post-mortem
Analysis. IET Gener Transm Distrib 11(13):3287–3298
4. Kapoor G (2018) Fault detection of phase to phase fault in series capacitor compensated six
phase transmission line using wavelet transform. Jordan J Elect Eng 4(3):151–164
5. Sahani M, Dash PK (2019) Fault location estimation for series-compensated double-circuit
transmission line using parameter optimized variational mode decomposition and weighted
P-norm random vector functional link network. Appl Soft Comput J (Elsevier), 1–18 (2019)
6. Sekar K, Mohanty MK (2018) Data mining-based high impedance fault detection using
mathematical morphology. Comput Elect Eng (Elsevier) 69:129–141
7. Kapoor G (2018) Six phase transmission line boundary protection using mathematical
morphology. In: Proceedings of the IEEE international conference on computing, power and
communication technologies (GUCON), pp 857–861, Greater Noida, India (2018)
8. Rathore B, Shaik AG (2018) Fault analysis using alienation technique for three-terminal trans-
mission line. In: Proceedings of the IEEE 2nd international conference on power, energy and
environment: towards smart technology (ICEPE), pp 1–6, Shillong, India (2018)
9. Wang L, Liu H, Dai LV, Liu Y (2018) Novel method for identifying fault location of mixed
lines. Energies 11(1529):1–19
10. Chaitanya BK, Yadav A, Pazoki M (2019) An improved differential protection scheme for
micro-grid using time frequency transform. Elect Power Energy Syste (Elsevier) 111:132–143
Chapter 8
Analysis of Voltage Sag and Swell
Problems Using Fuzzy Logic for Power
Quality Progress in Reliable Power
System
Ankit Tandon and Amit Singhal
1 Introduction
The electricity generation from conventional and non-conventional sources, its

transmission from DC or AC system, its distribution to domestic and commercial
consumers, and its utilization from human being and for industrial requirements
makes a power system network. In a smart power system, the difficulties during this
complete process are solved in a smart manner. Such type of disturbances includes
mainly VAR controlling and power quality issues. Managerial controlling from load
dispatch centers makes a power system better. Additionally, the fuzzy-based expert
systems can be implemented so as to make the existing system smarter with higher
accuracy.
The problems due to power quality and problems arising are penalty of more
utilization of solid-state switching devices, nonlinear load, electronic load, and
switching load. The arrival of high rating semiconductor switches in distribution
and transfer leaves current to be non-sinusoidal [1]. Electronic load causes voltage
distortion and harmonic distortion. Power quality problems create onset of systems,
sensitive equipment, data loss in the commuter, and MAL functioning in the memory,
like computers, programmable logic controller, the protective and relay apparatus [1].
A voltages sag or voltage dip is a short duration decrease in rms voltage. It is
caused by short circuit phenomena, overload, and starting of electric motors. If the
value of RMS voltage reduces from 10 to 90 percent of the nominal voltage for a
cycle to a minute duration, then phenomena of voltage sag is produced. Voltage sag
is termed as sustained sag if this low voltage remains in the system for a longer
duration up to a minute. Voltage swell is the contradictory of voltage sag. It is the
phenomena of an increase in the level of voltage which occurs due to turn off of a
A. Tandon (B) · A. Singhal

Department of Electrical Engineering, JIET, Jodhpur, India
e-mail: ankit.tandon@jietjodhpur.ac.in

https://doi.org/10.1007/978-981-15-8820-4_8
86 A. Tandon and A. Singhal
heavy load in an electrical power system. Similar to sag, swell happens for a very
short duration, and if it is of longer duration, then termed as sustained swell.
Voltage sag mainly increases the problems of power quality. It affects the
electricity distribution system and various industries which causes very high loss
of power. Short duration voltage dips can cause the whole industry out from their
normal operating condition. Generally, it is considered that the voltage dip as a source
10–90%. The main reasons include short circuit condition and breakdown voltage
dips, lightning phenomena, and power surges. In overhead transmission and distri-
bution systems, lightning phenomena is the main reason for producing voltage sag,
with an approximate occurrence of 50% of total cases [2]. The following pie chart
shows the main reasons for poor power quality and their approximate percentage.
To minimize conflict of power quality in electrical devices, some well-organized

detection and categorization techniques are necessary in the electrical power systems.
Types of power quality disturbances which are based on visual check of waveforms
by human being operators is a time consuming and laborious task. Also, it’s not
always necessary that we can remove some important information from simple visual
inspection [2]. The categorization of Power quality disturbances in an electrical power
system has now become a very important job for proper development and design of
preventive and corrective actions.
In the mid of 1950s, artificial intelligence has emerged as a new stream of
computer science engineering. There are many powerful tools produced by artificial
intelligence, which are used practically in different engineering streams so as to
resolve those complex engineering problems which require human intelligence. In
this paper, fuzzy logic tool is used for the analysis of disturbances like voltage sag
and swell [2, 3].
Fuzzy Controller
Fuzzy controllers are very simple and it is a new addition to control theory. They
consist of an input stage, a processing stage, and an output stage. Its design philosophy
differs from all previous methods by containing dexterous knowledge in controller
design. These FLCs are attractive choices where specific mathematical modeling
formulas are not possible. It has good control robustness as compared to tradi-
tional control scheme. It modifies a dexterous knowledge-based control strategy
into automatic control strategy in essence [3].
8 Analysis of Voltage Sag and Swell Problems … 87
Knowledge Base
Rules
Crisp Input Crisp Output
Fuzzifier Defuzzifier
x y
Inference
Fuzzy input Fuzzy output
sets sets
Fig. 1 Architecture of a Fuzzy Controller
A Mamdani-based fuzzy controller is used with maximum-minimum interference

method. It performs the function in the following 4 stages which is shown in the
following Fig. 1.
(1) Fuzzification process

(2) Rule base
(3) Inference
(4) Defuzzification process.
Fuzzification is the procedure of changing real scalar value into a fuzzy value
which is achieved with the different types of fuzzifiers. (Membership functions).
Rule base scheme is used as a way to store and manipulate knowledge to
understand the information in a useful manner. They are used in AI.
FIS is a system that uses the theory of fuzzy set for mapping of inputs. The input
is features while the output is classes of fuzzy classification.
Defuzzification is the process of getting a scientific result in the Crisp logic,
available fuzzy sets, and related membership degrees. So, the process of mapping of
fuzzy to crisp set is called as Defuzzification.
The planned fuzzy-based expert system is developed to classify short voltage
disturbances which can be defined as instantaneous and temporary sag, swell, and
interruption, which are shown in Fig. 2 [3]. In the study, the disturbance data are
obtained from PQ monitoring in which the monitoring software by default has three
different sampling frequencies of 0.4 kHz (128 cycle), 1.6 kHz (32 cycle), and 6.4 kHz
(8 cycles) and each frame has 1024 samples.
To process the raw disturbance data so as to remove features of the various distur-
bances, preprocessing of the disturbance signals is required. Initially, fast Fourier
transform analysis is used to separate the 8, 32, and 128 cycle waveforms. Then
root mean square (rms) method is applied by first approximating the fundamental
frequency profile of actual voltage waveform and determining the maximum and
minimum voltage magnitudes. An advantage of this method is its simplicity, fast
calculation, and less requirement of memory because rms voltage can be stored
periodically instead of per sample [4].
Power Quality Actual Data
FFT
8 Cycles 32 Cycles 128 Cycles
Obtaining rms voltage Obtaining rms voltage Obtaining rms voltage

& & &
Extract fuzzy inputs Extract fuzzy inputs Extract fuzzy inputs
Fuzzy logic to identifying

& classifying disturbances
Instan. Sag Momen. Sag Instan. Momen.

Non- Swell Interruption Interruption
Non-Sag Instan. Swell Momen. Swell Non-Interruption
Fig. 2 Design of the proposed fuzzy-expert system
2 Fuzzy Logic Inputs and Outputs
The Mamdani-type fuzzy system with five inputs and three outputs has been
considered in the proposed fuzzy-expert system. The five inputs include maximum
voltage in p.u (Max-V), sag duration in second (Sag Durat), swell duration in second
(Swell Durat), transient duration in second (Tran Durat), and minimum voltage in
p.u (Min-V) [5]. The maximum voltage has been chosen as a fuzzy input variable
so as to classify instantaneous and momentary swell whereas the minimum voltage
is for differentiating between interruption and voltage sag.
The sag duration is used as an input variable for classifying instantaneous and
momentary sag whereas the swell duration is used to classify swell disturbance to
instantaneous and momentary swell. The transient duration is chosen as a fuzzy input
for distinguishing between instantaneous swell and impulsive transient disturbance
[5]. The three FL outputs are Output1, Output2, and Output3 in which Output1
is designated for classifying instantaneous sag, non-sag, and momentary sag,
Output2 for classifying instantaneous swell, non-swell, and momentary swell, and
Output3 for classifying instantaneous interruption, non-interruption, and momentary
interruption [6].
3 Membership Functions
The input and output variables are represented by the most common shape of member-
ship functions which are either in trapezoidal or triangular forms and bell curves are
also used but the shape is less important than no. of curves. The range of input
variables and thresholds are chosen in accordance with the respective disturbance
definition as defined in the IEEE Std. 1159-1995. The following Tables 1 and 2
represent fuzzy sets for Input and Output Variables [7, 8].
The membership functions defined for the five input variables are as shown in
Figs. 3, 4, 5, 6, and 7. The output variables are defined by three membership functions
as shown in Figs. 8, 9 and 10.
If-Then rules (30) have been generated for classifying sag, swell, and interruption
disturbances. These rules are represented in the following form: IF premise THEN
consequent.
Table 1 Fuzzy sets defined for the input variables

Membership Input 1: Input 2: Sag Input 3: Swell Input 4: Input 5:
Function Maximum duration (s) Duration (s) Transient Absolute
voltage (p.u.) Duration (s) Minimum
voltage (p.u.)
1. L ESH ESH ESH VL
Low Extremely Extremely Extremely Very Low
short short short
2. M VSH VSH SH L
Medium Very Short Very Short Short Low
3. H SH SH SH M
High Short Short Short Medium
4. VH M M – H
Very High Medium Medium High
5. EH – – – –
Extremely
High
Table 2 Fuzzy sets defined

Membership Output 1 Output 2 Output 3
for the output variables
function
1 I sag I swell I interrupt
Instantaneous Instantaneous Instantaneous
sag swell interruption
2 N sag N swell N interrupt
Non sag Non swell Non
interruption
3 M sag M swell M interrupt
Momentary Momentary Momentary
sag swell interruption
Fig. 3 Maximum voltage input membership function
Fig. 4 Sag duration input membership function
Fig. 5 Swell duration input membership functions

Fig. 6 Transient duration input membership functions
Fig. 7 Absolute minimum voltage input membership function
Fig. 8 Output1 membership function

Example of the generated rules for identifying sag, swell, and interruption and
classifying them to instantaneous, momentary, and non-sag, swell and interruption
are given as follows:
(1) If (Max-V is L) and (Sag Durat is SH) and (Swell Durat is ESH) and (Tran
Durat is ESH) and (Min-V is L) then (Output1 is I sag) (Output2 is N swell) (Output3
is N interrupt). etc.
4 Conclusion
During the contingency analysis of an electrical power system, the variation in the
level of voltage and current are the two important issues that affects the reliability a
lot. In this paper, a fuzzy logic-based system is developed so as to analyze such types
of disturbances like sag, swell, transient signal, etc. If such types of disturbances
are detected and corrected, the reliability of existing power system network will
increase automatically. Five fuzzy inputs and three outputs are used with If-Then
rules in the fuzzy inference system (FIS). In this, a technique based on FFT with
RMS average method is used to describe the features of many kinds of disturbances.
The disturbances in the voltage are applied to the proposed fuzzy system to test the
accuracy, sag, swell, and other disturbances. Hence, better results are obtained with
the proposed fuzzy-based expert system that confirms the power quality progress
and makes the power system reliable.
References
1. “Power quality issues standards &guide lines”, IEEE, Vol 32, May 96
2. Ebron S, Lubkeman D, White M (1990) A neural network power distribution feeders. IEEE
Trans Power Deliver 5(2):905–914
3. Sultan AF, Swift GW, Fedirchu DJ (1992) Detection of high impedance arcing faults using a
multi-layer perceptron. IEEE Trans Power Delivery 7(4):1871–1877
4. Amouzad Mahdiraji, G, Mohamed A (2006) A fuzzy-expert system for classification of Short
duration voltage disturbances, Jurnal Teknologi, 45(D) Dis. 2006: 41–57, © Universiti Teknologi
Malaysia
5. Liao Y, Lee JB (2004) A Fuzzy-expert system for classifying power quality disturbances. J Elect
Power Energy Syst 26:199–205
6. Classification of Power Disturbances using Fuzzy LogicBoris Bizjak, Peter Planins University
of Maribor - FERI, Smetanova 17, Maribor, SI - 2000, Slovenia Giang Z.L., 2004
7. Gaouda AM, Kanoun SH, Salama MMA, Chikhani AY (2002) Pattern recognition application
for power system disturbance classification. IEEE Trans Power Deliver 17:677–683
8. Dash PK, Mishra S, Salama MMA, Liew AC (2000) Classification of power system disturbances
using a fuzzy expert system and a Fourier linear combiner. IEEE Trans Power Deliver 15:472–477
Chapter 9
Vehicle Detection and Its Speed
Measurement
Morium Akter, Jannatul Ferdous, Mahmuda Najnin Eva,

Sumaita Binte Shorif , Sk. Fahmida Islam,
and Mohammad Shorif Uddin
1 Introduction
In Bangladesh, road accidents are unfortunately increasing in an alarming rate day

by day. For example, in the year 2018 around 4317 dangerous accidents occurred in
Bangladesh which caused 10828 people are injured and 4580 people died according
to the statistics of road accidents in Bangladesh [1]. The most common reasons
for these road accidents are reckless driving, unlicensed and incompetent drivers,
distracted driving, drunk driving, driver’s unconsciousness, overspeed, and many
more [2]. However, most of these road accidents are caused due to overspeed of the
vehicles. Automated overspeeded vehicle detection and controlling may solve this
problem.
Some overspeed detection systems have been developed [3–9]. Ushakiranmai and
Thirumal [3] developed a system for vehicle speed measurement in traffic junctions
using magnetic sensors. The system consists of wireless anisotropic magnetic devices
M. Akter (B) · J. Ferdous · M. N. Eva · S. B. Shorif · Sk. Fahmida Islam · M. S. Uddin

Department of Computer Science and Engineering, Jahangirnagar University, Savar, Dhaka,
Bangladesh
e-mail: ecs_morium@yahoo.com
J. Ferdous
e-mail: polyjannat040@gmail.com
M. N. Eva
e-mail: mneva2015@gmail.com
S. B. Shorif
e-mail: sumaita.bs@gmail.com
Sk. Fahmida Islam
e-mail: jharnaju@yahoo.com
M. S. Uddin
e-mail: shorifuddin@gmail.com

https://doi.org/10.1007/978-981-15-8820-4_9
96 M. Akter et al.
which are placed next to the roadway and measure traffic in the immediate adjacent
lane. The accuracy of the system is good, but error is somehow higher. Shedbalkar
et al. [4] described a technique for vehicle speed estimation using permanent magnet
synchronous motors based on extended Kalman filter. However, the detection rate as
well as vehicle speed is not very good.
Benjamin et al. [5] developed a deep neural network-based method for car speed
estimation which gives 91% accuracy. But their developed system is slow. Abbas
[6] developed a system which estimates the vehicle speed using image processing.
The system performance suffers from poor lighting or bad weather. The system also
requires high performance signal processing hardware, large amount of memory,
and high bandwidth data links. Arash Gholami et al. [7] developed a system for the
detection of vehicle speed using CVS method, which works accurately in measuring
speed, but the system is sensitive to the thresholds. Budi Setiyono et al. [8] proposed
a Gaussian mixture model to detect vehicle speed using a sequence of images. The
speed is estimated by computing the movement of centroid in the sequence of frames.
Haque et al. [9] developed a primitive velocity estimation method by three-frame
differencing method.
The above methods are somehow expensive and their performances not up-to the
mark. Therefore, in this paper we are trying to develop a simple and cost-effective
complete system. Our system uses highly enhanced open source TensorFlow [10]
machine learning library object detection API (Application Program Interface) to
detect a vehicle and estimate its speed as well as overspeeded situation. The Tensor-
Flow models have different pre-trained models for various machine learning tasks
and their one important resource is the object detection API, which uses neural
network-based SSD (Single-Shot Detector) algorithm [11].
The rest of the paper is organized as follows. Section 2 deals with designing of
the system. Section 3 presents the experimental results and finally Sect. 4 concludes
the paper.
2 System Design
Figure 1 describes the flow diagram of our system. At first, the acquired input video
is fragmented into sequential frames by OpenCV. We marked a fixed line, known
as ROI (Region of Interest) line from which we perform our system computation.
Then the ROI line approaching vehicle is detected by using TensorFlow API. There
are many models which can be used to train our detection system, but we have
selected “ssd_mobilenet_v1_coco_2017_11_17”, as it consumes less memory. The
API model creates a bounding box including ROI (region of interest) line after
detecting the location of the vehicle. After that, it measures the speed of each
vehicle, and, also determines whether the vehicle is running at overspeed or not.
The speed is calculated in our system using the usual velocity formula, as there is a
pre-mapping between pixel displacement in image domain and real displacement in
vehicle domain.
9 Vehicle Detection and Its Speed Measurement 97
Fig. 1 Flow diagram of the

system Read Input Video Frame by
Frame (Using OpenCV)
TensorFlow-Based
Vehicle Detection
Encircle the vehicle with

bounding box
Calculate speed from

image pixel location
Speed > Speed limit
Yes No
Overspeed Normal Speed
Speed = distance/time (1)
We then check this estimated speed crossed the speed limit or not along with
annotation. The annotation is performed by using “LabelImg” tool from GitHub
[12] and the annotated data are saved as XML/CSV file.
3 Experimental Results and Analysis
We have tested the system with several sample videos taken by a mobile phone
(OPPO F1) camera from real scenes. If a vehicle’s speed is more than 50 km/hr,
then we have taken it as “overspeed”. Figure 2 shows an experimental scene of a
vehicle where we estimated its speed. The estimated speed is found normal. Figure 3
shows another experimental scene of a vehicle where we estimated its speed crossed
the speed limit. Our method failed to estimate speed if the scene contains multiple
vehicles. Figure 4 shows such a scene, where our system could not estimate vehicle
velocity. We have compared the estimated speed with the correct speed of a vehicle
as shown in Table 1. Here, we have taken the measured speed by “Speed Gun App”
98 M. Akter et al.
Fig. 2 Vehicle detection and speed estimation (normal speed)
Fig. 3 Vehicle detection and speed estimation (overspeed)
as correct speed. Our system’s speed measurement is almost matched with the speed
of the “Speed Gun App”. However, we have seen that error is higher for higher speed.
4 Conclusions and Future Works
This paper presents a TensorFlow-based system that detects the vehicle in a scene
and estimates its speed for the detection of overspeeded vehicle. The obtained results
9 Vehicle Detection and Its Speed Measurement 99
Fig. 4 The system could not estimate the vehicle speed due to multiple vehicle
Table 1 Comparison of real

Real speed (km/h) Obtained speed (km/h) Accuracy (%)
speed with obtained speed
23.6 21.97 93.00
58.01 53.57 92.30
49.80 45.88 92.14
62.80 68.77 91.30
106.50 89.92 84.40
are quite satisfactory. This system will help in reducing road accidents by detecting
overspeeded vehicles. The main limitation of our system is that it cannot estimate
the speed if the scene contains mutiple vehicles. So, our system should be placed in
a fixed location to estimate the vehicle speed, so that the scene should contain only
one vehicle. As our future work, we expect to develop a system which can estimate
the speed of multiple vehicles in a single shot.
References
1. Tribune D (2018) saw surge in road accidents. https://www.dhakatribune.com/bangladesh/nat

ion/2019/01/16/2018-saw-surge-in-road-accidents
2. Sivaraman S, Manubhai Trivedi M (2013) Looking at vehicles on the road: a survey of vision-
based vehicle detection, tracking, and behavior analysis. IEEE transactions on intelligent
transportation systems, vol 14, no 4, December 2013, pp 1773–1795
3. Ushakiranmai M, Sai Thirumal G (2015) Design Portable sensors system for vehicle counting,
classification, and speed measurement in traffic junctions. Int J Prof Eng Stud V(5):174–179
100 M. Akter et al.
4. Shedbalkar K, Dhamangaonkar AP, Walchand ABP (2012) Speed estimation using extended
Kalman filter for PMSM. In: 2012 international conference on emerging trends in electrical
engineering and energy management (ICETEEEM), Chennai, India, 13–15 December 2012,
pp 433–435, IEEE Explore
5. Penchas B, Bell T, Monteiro M (2019) A deep learning approach to vehicle speed estimation.
http://cs229.stanford.edu/proj2017/final-reports/5244226.pdf. Accessed 30 July 2019
6. Awni Abbass A (2010) Estimating vehicle speed using image processing. AL-Mansour J
14:127–141
7. Gholami Rad A, Dehghani A, Rehan Karim M (2010) Vehicle speed detection in video image
sequences using CVS method. Int J Physical Sci 5(17)–2555–2563
8. Setiyono B, Ratna Sulistyaningrum D, Soetrisno, Fajriyah F, Wahyu Wicaksono D. Vehicle
speed detection based on Gaussian mixture model using sequential of images. J Phys: Conf Ser
890(1):012144. IOP Publishing. https://iopscience.iop.org/article/10.1088/1742-6596/890/1/
012144/pdf
9. Reduanul Haque M, Golam Moazzam Md, Islam S, Das R, Shorif Uddin M (2016) Vehicle
speed determination from video streams using image processing. In: International workshop on
computational intelligence (IWCI), 12–13 December 2016, Dhaka, Bangladesh, pp 252–255,
(IEEE Explore)
10. Francis J (2019) Object detection with tensor flow. https://www.oreilly.com/ideas/object-det
ection-with-tensorflow. Accessed 30 July 2019
11. Liu W, Anguelov D, Erhan D, Szegedy C, Reed S, Fu C-Y, Berg AC (2016) SSD: single shot
multi box detector. ECCV 2016, Amsterdam, Netherlands, October 11–14, 2016, Lecture Notes
in Computer Science, vol 9905, Springer, pp 21–37, 2016
12. A Graphical Image Annotation Tool—“LabelImg,”. https://github.com/tzutalin/labelImg
Chapter 10
Fuel Cells as Naval Prime Movers:
Feasibility, Advances and Implications
Akshat Mathur and Sushma Dave
1 Introduction
Traditionally, the guidelines for the selection of naval prime movers focused on the
demands for greater speed, better shock absorption, low noise generation and low
infrared signature; but modern warships require a more versatile package which
should be flexible enough for broadly varying operational requirements. Modern
propulsion systems are expected to operate in a greater range of speed with longer
durations of deployment and cleaner emissions [1] while aiming to operate with a
numerically inferior manpower compliment than ever before.
Fuel cells inherently offer reduced maintenance costs, emissions and infrared
signatures, acoustic signatures, radar cross-section and increased ship survivability
due to distributed power reduction and greater flexibility of design and operation
due to their modular nature [2]. Fuel cells produce such low levels of noise that the
fuel cell cabin on board a ship is expected to have noise levels similar to that of
any regular office [3]. Thus, they promise to be a more promising approach towards
a clean and efficient power solution as compared to other power sources or prime
movers in both stationary and mobile applications. However, their foray into naval
propulsion systems is relatively new and is still in its infancy in both maturity and
deployment.
The first fuel cells to enter production for naval propulsion as prime movers are
Siemens’ PEM fuel cells employed by German-Italian Type 212A submarines along
with its export derivative: the Type 214 as a part of their Air Independent Propulsion
mechanism. These boats are in service with various navies across the globe including
the German, Italian, Hellenic (Greek), Portuguese and South Korean navies.
A. Mathur (B) · S. Dave

Jodhpur Institute of Engineering and Technology, Jodhpur, RJ, India
e-mail: mathur.akshat@outlook.com

https://doi.org/10.1007/978-981-15-8820-4_10
102 A. Mathur and S. Dave
Furthermore, on the economic front, the U.S. Navy estimated that if a ship service
fuel cell plant, having an efficiency of 37–52%, replaces the existing gas turbines
on board a DDG 51 Arleigh Burke class destroyer, then, considering a 3,000-h time
window, it would consume only a paltry 33% of the fuel relative to what would have
otherwise been consumed by the warship. This, the U.S. Navy estimated, could save
them more than $1 million per ship per year in the form of ship service fuel costs [2].
Despite the proliferation of fuel cell-powered Air Independent Propulsion
equipped submarines around the world in the last few years, the technology has
not gained much turf in the realm of surface vessels: especially naval. In [4], Jing
Sun, et al. note that there is a lack of published studies focusing on larger vessels
with greater power requirements. They propose a notional Solid Oxide Fuel Cell
module and analyse the fuel saving and machinery arrangements by using a hybrid
SOFC-gas turbine module for a U.S. Navy Sealift vessel. They concluded the same
benefits of employing fuel cell prime movers but also noted that large technology
gaps exist and such power density levels, as they have considered in their study, have
not been yet achieved by modern state-of-the-art fuel cell technologies.
However, there have been many attempts to construct viable fuel cell modules;
the developments and deployments of which are discussed further in this paper.
2 Naval Fuel Cell Technologies
2.1 The Notional 5 MW SOFC Unit
Fuel cell technology has been making rapid strides with the worldwide effort to
develop cleaner and more efficient power sources. A study conducted by Jing Sung,
et al. identified the baseline desirable parameters of a marine fuel cell prime mover.
Four such notional fuel cell units were coupled with a gas turbine to obtain a total
output power of 24 MWe [4]. Principal parameters of this hybrid SOFC-GT system
are listed in Tables 1 and 2.
Using this system, a mission endurance analysis was performed on a U.S. Sealift
Command’s Roll on–Roll off (Ro-Ro) cargo vessel, and the mission endurance was
determined by the amount of distance travelled by the ship with 90% of the full fuel
capacity at the peak speed of 24 knots [4].
Table 1 Principal SOFC

SOFC Parameter Magnitude
parameters of the hybrid
SOFC-GT system [4] Output Power (MWe ) 5
Efficiency (%) 50
Fuel Utilisation (%) 85
Air Utilisation (%) 25
Air/fuel inlet temperature (°C) 600
10 Fuel Cells as Naval Prime Movers: Feasibility … 103
Table 2 Principal GT
GT Parameter Magnitude
parameters of the hybrid
SOFC-GT system [4] Compressor Efficiency (%) 5
Turbine Efficiency (%) 50
Pressure Ratio 85
Generator Power (MWe ) 25
In this analysis, the fuel cell-powered ship was found to have an endurance advan-
tage of 25% over diesel systems and 57% over gas turbine systems while consuming
20% and 50% less fuel than the said prime movers, respectively. Furthermore,
amongst all the conditions evaluated, the power output of the combined system never
fluctuated by more than 2% which indicated a reliable design and arrangement [4].
However, this analysis was performed with an auxiliary vessel of the U.S. Sealift
Command and, therefore, does not adequately reflect the demands of a frontline
warship. Also, it was acknowledged by the authors that present technology does not
allow us to field such systems, however, they predicted that they can be developed
in the near future, i.e. within a span of 15–20 years [4].
2.2 Siemens SINAVY PEM Fuel Cells
Siemens offers two kinds of SINAVY Proton Exchange Membrane fuel cells: the
FCM 34 and the FCM 120. The FCM 34, developed in 1984 at the request of the
German Ministry of Defence, is employed by Type 212A submarines which are
in service with the German and the Italian navies. The FCM 120, developed at
a later date, is deployed in Type 214 submarines which are Type 212A’s export
variant currently in service with various navies across the globe including the
Hellenic (Greek), Portuguese and South Korean navies [5]. Table 3 lists the technical
specifications of FCM 34 and FCM 120 SINAVY fuel cells.
The voltage of SINAVY PEM Fuel Cells with respect to the operating time is
quite stable. The degradation rate of the system is less than 2 µV/h per cell for a
FCM 34 module and significantly lower than the same for a FCM 120 module [5].
This property is particularly desirable as naval vessels have stringent operational
requirements wherein all the deployed equipment is expected to adhere to their
specifications for the duration of a reasonably long service life along with extended
operating time durations.
These fuel cells are employed by the aforementioned submarines as a part of
their Air Independent Propulsion system which allows them to say submerged
for relatively longer durations than what their diesel engines would have allowed
for. However, under the fuel cells’ power, the boats can only cruise at a speed of
approximately 5 knots (submerged).
In comparison to this, the cruising speed of naval surface escort ships, which
tend to be of significantly greater displacements and thus requiring greater power
Table 3 Specifications of SINAVY FCM 34 and FCM 120 SINAVY fuel cell modules [5]
Technical data FCM 34 FCM 120
Rated power (kWe ) Approximately 34 Approximately 120
Voltage range (VDC ) Approximately 54 Approximately 215
Efficiency at rated load (%) >58 >53
Efficiency at 20% load (%) Approximately 71 Approximately 68
Operating temperature (°C) 70–75 Approximately 70
H2 pressure (bar) 2.3 (absolute) 2.4 (absolute)
O2 pressure (bar) 2.6 (absolute) 2.7 (absolute)
Dimensions (cm) H = 47 H = 50
W = 47 W = 53
L = 143 L = 176
Weight (kg) (without module electronics) 630 930
Maximum applied current in continuous operation (A) 650 560
Technical data FCM 34 FCM 120
densities than their sub-surface counterparts, is about 10–15 knots while peak power
allows them to achieve 25–30 knots conventionally in the open seas. These require-
ments pose a challenge for fuel cells as significant technology gaps exist which need
bridging if such systems are to be successfully incorporated in surface vessels as
prime movers.
However, due to inherent flexibilities offered by fuel cells, it is possible to refit
operational submarines with SINAVY PEM fuel cell modules and, thus, extend AIP
capabilities to those boats. Furthermore, as shipboard reformer technology advances
suitably, fuel cell technology’s field of application can be expanded and it may
become the preferred prime mover for sub-surface and surface vessels [5].
2.3 Fuel Cell Energy Inc. Direct Carbonate Fuel Cell
In 1997, the Office of Naval Research of the United States sponsored a three-
phase project to demonstrate the viability of commercial fuel cell technology in
ship service applications on board its warships whose conceptual design criteria are
listed in Table 4. By 2003, FuelCell Inc. completed the conceptual design and critical
component testing of a 2.5 MW fuel cell module for future surface combatants. To
achieve this, FuelCell Inc. used its Direct carbonate Fuel Cell (DFC) technology
which provided thermal efficiencies in excess of 50% [6].
The device is fuelled by reforming methane and steam to produce carbon dioxide
and hydrogen which is then fed to the anode of the cell. The power-producing anode
reaction converts hydrogen to steam which allows reforming reaction to tend towards
the product side at a relatively low temperature of 650 °C by continuously shifting the
Table 4 Fuel cell energy Inc.

Parameter Magnitude
direct carbonate fuel cell
specifications [5] Power (MWe ) 2.5
Voltage range (VAC ) 450
Efficiency at 50% load (%) 40
Spatial allowance (m3 /kWe ) 0.057
Weight allowance (kg/kWe ) 18
Production cost ($/kWe ) 1,500
reaction equilibrium to obtain the desired products. Such a commercial fuel cell stack
typically comprises 350 individual cells arranged in a cuboidal fashion measuring
1.2 m × 0.6 m [6].
PEM fuel cells, in general, require storage of highly flammable and dangerous
hydrogen gas which can be a potential safety concern on board any ship. An error,
accident or battle damage (in case of warships) may result in uncontrolled discharge
of the stored hydrogen gas which can be a significant health hazard on board any
vessel. Therefore, internal reforming in the direct carbonate fuel cell at a relatively
moderate temperature is a desirable capability on board naval vessels as it only
requires the storage of relatively safer hydrocarbon fuels.
The efficiency, size and weight constraints, as indicated in Table 4, were deter-
mined by the U.S. Office of Naval Research so as to be comparable with marine diesel
engines and offshore power generators. The production cost of the system was deter-
mined as such to be “somewhat” lesser than that of large diesel engines [7]. These
parameters, therefore, indicate the absolute minimum respective magnitudes that
ship bound fuel cell technology should attain to be operationally and economically
feasible as compared to conventional diesel prime movers.
3 Integration
Conventionally, naval propulsion systems comprise internal combustion engines, viz.

gas turbines and diesel engines. These systems are employed in a combination in the
propulsion system with different prime movers often being employed for different
purposes such as propulsion, services and auxiliary demands. The integration of
notional fuel cells in such an arrangement is outlined in [4] and is illustrated in
Fig. 1.
In modern propulsion systems, however, the concept of Integrated Full Electric
Propulsion (IFEP) is gaining turf. IFEP is defined as the use of a common power
system for both propulsion and ship’s services. In such an arrangement, efficient
operation is obtained by the use of the minimum number of prime movers necessary
to meet the load which all run at their optimum efficiency [8].
It is lucrative for applications in naval propulsion because of the inherent benefits
that it carries. A lesser number of prime movers would imply lesser sources of noise
Fig. 1 Combined Fuel Cell and Gas Turbine arrangement for marine applications [4]
generation which is a very desirable outcome for any warship. Furthermore, such a
system will be less maintenance intensive and would, thus, require a lesser manpower
compliment which would result in reductions in life cycle costs [9]. The first warships
to wield an Integrated Full Electric Propulsion system are the Royal Navy’s Type 45
air defence destroyers. Their IEP system is illustrated in Fig. 2 [9]. This concept is
also known as that of an All Electric Ship (AES).
In [10], the authors consider that in an AES, power can be provided by the means
of diesel generator motors through transformers or by battery banks connected to
the motors and evaluate three “options” of integrating hydrogen-oxygen fuel cells
on board ships. The outcome of the analysis of these 3 arrangements, as determined
by the authors, is presented in Table 5.
Fig. 2 Electric Propulsion System of Type 45 destroyers [9]

Table 5 Comparison of options for fuel cell integration on board ships [10]
Parameter Option 1 Option 2 Option 3
Advantage(s) Less dependence on fuel Without batteries or Power system totally
cells battery charge system operated with
renewable energies
Disadvantage(s) Fuel cells in parallel The diesel generator is Requires 8 fuel cell
with batteries; the diesel still in the system modules and a 12 V
generator is still in the DC-DC converter
system
Complexity of the High Low Low
control system
Space Not sufficient Not sufficient Adequate
Emissions High High Low
Harmonic High Low Low
distortions
It is noted that diesel generators pose the problem of limited energy production
and are sensitive to harmonic current consumption. Furthermore, the batteries have
the disadvantage of relatively longer charging time as compared to the subsequent
duration of operation. Also, the impedance of the diesel generator in combination
with the non-sinusoidal current feedback of the battery chargers produces voltage
distortion that limits the functionality of the power system [10].
Option 1 comprised of an arrangement in which fuel cells are arranged in parallel
with batteries. This arrangement adversely affected the system’s response time due
to the different sources. Currently, fuel cells are not directly controlled causing disar-
rangement between the fuel cell and battery impedances, which makes it difficult to
implement power systems that have fuel cells in parallel with the batteries [10].
Furthermore, an elaborate control system would be required to match the fuel
cell’s voltage, which varies from 440 to 800 V, to that of the batteries. However, the
space that would be gained by eliminating a series of batteries to house the fuel cell
modules and power conditioning system would be insufficient for the said task [10].
In option 2, the batteries and the continuous current charge mechanism were
completely eliminated from the system. This option benefits by the removal of the
limited use time and the presence of harmonic distortions due to these eliminations.
Also, an increase in the overall power density of the system is thus achieved. However,
the space constraints still do not allow for the accommodation of the fuel cells and
their fuel and cooling systems in the space created by the elimination of battery stacks
[10].
In option 3, the entire power system of the vessel is replaced by fuel cells and a
DC-DC converter. Such an arrangement will be able to run completely on renewable
energy and would operate below 80 °C making it “highly reliable”. Furthermore,
such an arrangement would be highly efficient with low emissions and would also
be able to avoid the problem of harmonic distortions [10].
For such an arrangement to succeed, fuel cell modules would need to be connected
in parallel with each other to satisfy the system’s needs. However, upon connecting
the fuel cells to a bus in parallel, a high possibility exists that the current flows back
into the fuel cell modules which can reduce their useful life and service efficiency
[10]. All of these analyses are summarised in Table 5.
4 Fuel
Like the prime movers, it has the potential to replace, fuel cells also have the ability to
utilise hydrocarbon fuels, amongst other kinds of fuels. However, since the produc-
tion of consumable energy, in the form of heat and electricity, only takes place
through electrochemical reactions, and not through combustion, as opposed to the
internal combustion engines in widespread use, emissions from fuel cells are mostly
negligible in nature. Table 6 summarises the various compounds which may be used
by a particular kind of cell with a potential for naval application as a fuel, and those
which may act as a poison for them beyond certain concentrations.
Alkaline fuel cells as in contrast to the listed kinds of cells, they necessarily
require pure hydrogen as a fuel for their operation and cannot operate on a reformed
fuel as noted in [12], which was a study conducted by Arctic Energies Ltd. for the
feasibility analysis of replacing the diesel-electric propulsion of a U.S. Coast Guard
ship which later resulted into the development of the (Molten Carbonate) Direct Fuel
Cell technology (2.3).
On board reforming of fuel is desirable for fuel cell operations as it avoids the
shipboard storage of hydrogen which is dangerously flammable. The same influenced
the Russian Navy’s AIP choices and, in the light of the records of frequent fires on-
board its ships, it chose not to develop fuel cells similar to the ones employed by
German submarines (2.2) for its new Lada class diesel-electric boats, but instead went
forward with developing technology that would “convert diesel fuel into hydrogen
for power” [13], i.e. reform it.
Table 6 Various gaseous compounds as fuel and poison for various kinds of fuel cells for potential
naval applications [11]
Gas Phosphoric Acid Molten Solid Oxide Fuel Proton Exchange
Fuel Cells Carbonate Fuel Cells Membrane Fuel
Cells Cells
H2 Fuel Fuel Fuel Poison (>10 ppm)
CO Poison (>0.5%) Fuel Fuel Diluent
CH4 Diluent Diluent Fuel Diluent
CO2 & H2 O Diluent Diluent Diluent Diluent
S as H2 S & COS Poison Poison Poison No studies
(>50 ppm) (>0.5 ppm) (>1.0 ppm)
In [3], the authors conduct a thermodynamic analysis of a 120 kWe diesel-fuelled

SOFC system and conceive it as an auxiliary engine while comparing it with a diesel-
electric generator set for surface ships. They report a reformer efficiency of 85%,
fuel cell efficiency of 68.47% and, thus, a net efficiency of 55.28% while using the
NATO F-76 marine diesel fuel.
Also, the higher operating temperatures of MCFCs and SOFCs, in conjunction
with their ability to use reformed fuels, can be particularly beneficial for warships
which have limited power supply as the waste heat from the said cell stacks can be
utilised for fuel reforming which also requires high temperatures to occur.
5 Conclusion
We discussed various aspects of deploying fuel cells as naval prime movers covering
the assessment of their feasibility, review of the recent advances and understanding
of their implications. Furthermore, their fuelling and integration within the existing
marine propulsion systems were discussed.
We conclude that fuel cells offer a promising option for the future of naval propul-
sion with numerous environmental, economic and combat benefits. However, it is
notable that no fuel cell systems are presently deployed as naval prime movers with
the notable exception of the German-Italian Type 212 submarines with their Air Inde-
pendent Propulsion systems. However, a significant number of studies and attempts
have been consistently made around the world which were discussed in the review.
It was observed that the principal hurdle in the path of realising fuel cells as naval
prime movers is their poor spatial efficiency and weight-to-power ratios (specific
power). Furthermore, certain types of fuel cells, such as the SOFC and MCFC,
which are gaining the turf for naval applications, operate at very high temperatures.
All the fuel cells discussed in this study prefer hydrogen as fuel which requires
its on-board storage. This poses significant risks in a marine environment due to
the dangerously flammable nature of hydrogen. It is, therefore, safer and cheaper
to employ shipboard fuel reforming technologies to produce the necessary fuel for
the fuel cells on board in real time from the regular marine diesel fuel which would
save considerable monetary resources in the terms of on board and off board storage,
handling, transportation and processing. With the notable exception of alkaline fuel
cells, most other fuel cells can comfortably operate on reformed fuels with both
SOFCs and MCFCs benefitting particularly from this arrangement.
Further innovations in the area, therefore, need to be directed towards improving
the spatial efficiency, the specific power and the reforming techniques for fuel cells
so that their numerous benefits, especially environmental, can be suitably exploited
for the better.
References
1. Ohmayer HF (2012) Propuslion system choices for modern naval vessels

2. O’Rourke R (2006) Navy ship propulsion technologies: options for reducing oil use -
Background for Congress
3. SINAVY PEM Fuel Cells: For Submarines. https://assets.new.siemens.com/siemens/assets/
public.1535009488.28615cde70250d0e81b68ba466bd77d7f5c68c73.sinavy-pem-fuel-cells.
pdf. Accessed 08 April 2019
4. Abens S, Ghezel-Ayagh H, Steinfeld G, Sanderson R (2000) Development of a Ship Service
Fuel Cell. AES 2000/All Electric Ships
5. Application of fuel cells in surface ships. https://www.osti.gov/etdeweb/servlets/purl/202
49928. Accessed 13 April 2019
6. Fuentes DI, De Lavalle PÃ©rez H (2015) Fuel cell integration on board surface ships. Ship Sci
Technol 8(6):19–28
7. Fuels for fuel cells, http://www.nfcrc.uci.edu/3/FUEL_CELL_INFORMATION/FCexplained/
Fuels.aspx. Accessed 27 April 2019
8. Feasibility Study of Repowering USCGC VINDICATOR (WMEC-3) With Modular Diesel
Fueled Direct Fuel Cells. https://apps.dtic.mil/dtic/tr/fulltext/u2/a337901.pdf. Accessed 27
April 2019
9. Why Russia’s New ‘Stealth’ Submarines Have a Big Problem. https://nationalinterest.org/
blog/why-russias-new-stealth-submarines-have-big-problem-22941?nopaging=1. Accessed
27 April 2019
10. Fuentes DI, De Lavalle PÃcrez H (2015) Ship science and technology
11. Fuels for fuel cells (National Fuel Cell Research Centre, University of California, Irvine), http://
www.nfcrc.uci.edu/3/FUEL_CELL_INFORMATION/FCexplained/Fuels.aspx. Accessed 27
Apr 2019
12. Kumm WH, Lisle Jr HL (1997) Feasibility study of repowering USCGC VINDICATOR
(WMEC-3) with modular diesel fueled direct fuel cells. Arctic Energies LTD. https://apps.
dtic.mil/dtic/tr/fulltext/u2/a337901.pdf. Accessed 27 Apr 2019
13. Roblin S (2017) Why russia’s new ‘Stealth’ submarines have a big problem. The
National Interest. https://nationalinterest.org/blog/why-russias-new-stealth-submarines-have-
big-problem-22941?nopaging=1. Accessed 27 Apr 2019
Chapter 11
Overview of Eco-Friendly Construction
Materials
Nidhi Sharma and Aashish Kumar Jha
1 Introduction
Central Europe reflects a prominent example of eco-materials in the form of the

Mostar Bridge being constructed. Suleman is a magnificent sixteenth century autho-
rized new bridge as a substitution for a wooden bridge. [1]. In Spain, according to the
Spanish power system advanced report in 2016, the author of the business group “Red
Electra de Espana” said that electricity generated from renewable sources was 41.1%
in 2016, compared to 58.9% from non-renewable sources. There has been a decrease
compared to the year 2015, in which 63.1% of electricity was generated through non-
renewable sources. The common and basic sources of non-renewable generation of
electricity is the nuclear power, the thermoelectric power, coal plants, and the plans
of combined cycles, manifested the excessive production, with the created 55,546
GWH, 37,038 GWH and 29,787 GWH, respectively, in 2016 [2]. The paramount
challenges today are the management of wastes, which is generated on a large scale
and greenhouse gas emission generated by energy generation, industrial produc-
tion, and consumption models, green technologies or ecofriendly production models
should be the basic pillars to develop a circular economy. All efforts must focus on
flourishing the technology of ecology that helps reduce environmental adverse effects
and also improve society’s expansion and create enhancement for its betterment.
Road has been the most useful infrastructure that resulted from transport activities
and has helped in stimulating human civilization and development. There has been
a long history of road construction in the twentieth century BCE, the Arab Republic
of Egypt constructed roads to transport large amounts of rocks from quarries to
N. Sharma (B) · A. K. Jha

Jodhpur Institute of Engineering and Technology, Jodhpur, Rajasthan, India
e-mail: nidhisharma9862@gmail.com
A. K. Jha
e-mail: aashishjha003@gmail.com

https://doi.org/10.1007/978-981-15-8820-4_11
112 N. Sharma and A. K. Jha
sites where the rocks were indulged in pyramid building and the Great Sphinx [3,
4]. In the reflecting history of Rome, people constructed an advanced road network
being centered in Rome had played a noteworthy role in the prosperity of the ancient
Roman. The Empire and the proverb had it: “all roads lead to Rome” [5]. In addition,
the “Silk Road”, which had been in operation since the second century BCE from
the thirteenth and fourteenth centuries BCE, is characterized by economic, cultural,
and technological exchanges between East and West of the Asian continent, had been
promoted a lot. It was great contribution to the economic development and social
progress of the world [6]. Then came the first study of the use and reuse of waste glass
as a building material in 1963 by Schmidt and Sia. They converted the discarded glass
into useable glass chips for making wall panels. Subsequently, extensive studies came
into being to reuse waste glass in construction materials as glass being the most versa-
tile materials across the globe owing to the excellent characteristics that are being
optically transparent, chemically inert, high intrinsic in strength and its permeability
is quite low. Glass powder is a shallow and promising eco-complementary cement
material. These are beneficial in geofilmers, forming eco-friendly cement, artificial
lightweight aggregates, and composite phase change materials. It has also helped in
sustainable development and betterment of industries by producing durable building
materials (Fig. 1).
In the twenty-first century being engulfed with global warming and climate change
era, the earth’s surface temperature is expected to rise approximately 1.1–6.4 °C; until
Fig. 1 SEM images representing a cement grain, b green glass (63–75 µm), c green glass (25–
38 µm), and d green glass (0–25 µm) [7]
11 Overview of Eco-Friendly Construction Materials 113
2100 [8]. And a 2 °C rise in the global temperature is noticeable because of the rise
in total emissions caused as a result of the combustion of fossil fuels to produce and
supply electricity to buildings [9].
Steered by the high energy demand owing to make use of electricity for both
heating and cooling purposes, the sector of the building has an extreme dependence
on fossil fuels by being its consumer and emits greenhouse gases extremely (GHGs)
[10]. The magnification and the development in space cooling in the building have
been expected to occur in developing economies due to the rise in the standard of
living [11, 12].
Many other pre-eminent eco-friendly materials can be manufactured and their
usage can be indulged more prominently in the constructional sector, such as AAC
Blocks, Low VOC paints, Low E glass windows, and also PPC and Bamboos being
discussed separately.
2 Objectives and Fact-Finding Views of the Contemporary

Study
The contemporaneous study presents the gist of the feasibility of eco-friendly mate-
rials. This study flourishes the remarkable extent of the theories and work that has
been handled and tackled in this field in different forms. The study was aimed at
manifesting the current knowledge as a base and has been attributed by featuring its
scope for further research in this field.
3 Coal Fly Ash
Coal fly ash is used as a raw material for the manufacture of two building materials,
which are clay bricks that include silica-calcareous non-foamed bricks. Clay bricks
are prepared with the help of commercial clay and varying waste ratios, taken as 0–
50%. At 10 mpa and is being molded and fired at 1000 °C. For the preparation of Non-
Fired clay bricks; basically, two materials are used in its formation, CFA and Geosilex
where Geosilex is a lime residue that is obtained from the acetylene industry waste. Its
mold is also prepared at 10 mpa and is cured at room temperature for 28 days. These
results indicate the concretization of CFA (20 wt%) that produces fired clay bricks
with the physical and mechanical properties similar to the control bricks without
any wastage. Because of the increase in open porosity; the mechanical properties are
reduced by 25–50%. Now, secondly in silica calcareous fired bricks, on a decrease in
fly ash content; a decrease in bulk density and water absorption is also observed. These
bricks contain 40–60 wt% of CFA had the highest compressive strength up to 46–43
mpa. Therefore, both kinds of bricks helped in optimal technological properties that
attain quality standards.
4 Glass Powder
Then came the first study of utilizing and reusing waste glass as construction materials
which were carried out by Schmidt and Saia in 1963. In construction industries,
recycling of waste glass helped in reducing the burden on landfills and contributed
to resource preservation and
carbon footprint reduction. Considering the case like Hong Kong; where out of
total waste glass bottles, 10% is recycled and 90% is landfilled. These discarded
glasses were converted into usable glass chips for utilizing it to construct the wall
panels. Due to the limitations originated from the researches, waste glass was quickly
converted into powder. This GP had angular shape and sharp edges with the particle
size below 600 µm. These include 70% SiO2 , 13% Na2 O, 10% CaO, and smooth
texture as compared to ordinary Portland cement grains. Therefore, being amorphous
in structure, it served as a pozzolan in cement concrete.
5 Eco-Friendly Road Construction Material
The study adopts the art of research on the applications, demonstrations, and chal-
lenges of six environmentally friendly functional road materials, which are noise-
reducing pavement materials, permeable asphalt concrete, low heat-absorbing pave-
ment materials, de-icing pavement materials, energy harvesting pavement materials,
and exhaust gas dissolved pavement materials. In relation to this study, we focus on
providing a combination of the latest relevant works of literature on the develop-
ment and design of environmentally friendly functional pavements and to promote
innovation in materials science and pavement design principles. Porous noise reduc-
tion asphalt concrete is used to reduce pavement noise which is based on the noise
reduction principle by pores. Hence, by considering and utilizing these adequate
pavement materials prevention of noise from horizontal propagation can be obtained
by the application of soundproof structures as well [13].
6 The Adoption of Eco-Block as a Green Construction

Material
In the Mauritius construction industry, cellular blocks were complete concrete

masonry units. These blocks are used in the construction of buildings due to its dura-
bility and structural strength. With the worldwide energy crisis and the constraint
of carbon footprints, stakeholders are moving toward ecological materials to reduce
energy consumption. Being a Paramount, it is a concrete block embedded with a
polystyrene core which means that the Eco-block has been introduced since 2015
with its thermal and acoustic insulation.
7 Portland Pozzolanic Cement [PPC]
It is particularly a contrast to ordinary Portland cement (OPC) which consists of a

mixture of pozzolanic materials known for their increased strength of concrete and
reduced cement content. Pozzolanic materials include fly ash, rice bran ash, volcanic
tuffs. After burning coal in thermal power plants, fly ash is obtained as a waste
product.
8 AAC Blocks
Autoclave Aerated Concrete Blocks are an alternative to red bricks. These are manu-
factured from cement, sand, lime, water, and additives. The use of fly ash as a
substitute instead of sand
being a fine aggregate is inculcated to produce AAC blocks. The characteris-
tics defining these blocks are: they are easy to handle, less fuel in transportation,
lightweight, decrease cost of construction by saving structural steel requirements for
high RCC framed structures.
9 Bamboo
Bamboos are the best eco-friendly material on the planet. They possess a very high
self-generation rate and grows up to 3 ft. in 24 h. These are perennial grasses that are
replicated after harvesting and therefore have a higher strength to weight ratio, higher
compressive strength, and exceptional durability than solid bricks therefore used for
flooring and cabinet purpose. Since ancient times it was used as a building medium
and has also gained popularity due to the lack of natural wood. They have five times
the strength of concrete and are lighter than steel. Therefore, it has been proved to be
good moisture resistance and scratch-resistant. They are available in various shades
and is cheaper than hardwood. Also, it enrolls a drawback that requires treatment of
insect resistance.
10 Low E Glass Windows
Low emissivity glass known as low E glass has an excellent insulation property.
There has been a special coating being applied to glass by which only the visible
light can pierce into the room and accords protection from UV and infrared rays.
11 Controlled Low-Strength Material
Modern hardening properties of controlled low-strength material (CLSM) which is

a clay-based controlled low strength material have been investigated according to
current studies. CLSM contains cementitious material like the stainless steel and this
form of CLSM fundamentally is being used as a trench-fill material for construction
purposes. Four levels of cement replacement (i.e., 0%, 10%, 20%, and 30%) are
characterized by four levels and three quantities of binder content (80-, 100-, and
130 kg/m3 ), Generated for investigation work. The modern properties of hardening
of this recommended CLSM have been inspected in the laboratory. Therefore, the
proposed CLSM can be employed for environmentally friendly trench fill based on
the result being shown. In the present day, CLSM have been commonly used as a
subgrade, granular compacting soil, and trench fills. As it successfully consumes a
gigantic quantity of waste materials that is combustion fly ash, rubber, foundry sand
so it can be used as an EFM. CLSM typically contains small amounts of Portland
cement, mixing water, and large amounts of supplementary fine aggregates, CLSM
has 8.3 MPa or less compressive strength and is higher than compacted soils [14].
Subsequently, recent studies have also described that the CLSM’s maximum strength
of up to about 1.4 MPa is suitable for almost all backfilling applications, with future
re-excavation by simple equipment expected and desired. [14, 15].
12 Eco-Friendly Polymer Composites for Green Packaging
Green packaging is based on biodegradable composite materials that have a unique

property and has gained attention in the current scenario in comparison with the
classical photochemical-based plastics [16]. In the wide range of applications, its
uses can be featured and it can be used in intelligent nano-food packaging [17], bio-
membranes for wastewater, and composting purposes. Thus, the prominent function
for package material is particularly to achieve the benefits of improved food quality
and also ensure safety with prolonging and increased shelf life. Figure 2 represents
the classification of green reinforcements.
13 Conclusion
These paper and examples explain that the conceptualization of “eco-materials”

provide an interesting technological solution to enormous manufacturing products
that prove to help in the establishment of sustainability. The expansion of the eco-
friendly sustainable materials will help to address pollution due to the issues raised
by the environment and participate in the different outcomes and solutions of social
appropriateness. The progress in these eco-friendly systems will help to lessen the
Fig. 2 Classification of nanomaterials based on their categories [16]
impacts of pollution efficiently. Also, the motive is to work on recyclable products

that help in inducing sustainable requirements and needs. Hence, the above materials
provide the prominent use of eco-friendly materials and thereby fabricate an advanced
and empirical country.
References
1. Mirzahosseini M, Riding KA (2015) Influence of different particle sizes on reactivity of finely

ground glass as supplementary cementitions material (SCM). Cement Concr Compos 56:95–
105
2. IPCC, Pachauri RK (Ed), Climate Change 2007: Contribution of … To the Fourth Assessment
Report of the Intergovernmental Panel on Climate Change, IPCC, Geneva, 2008, p 4 Synthesis
report: [a report of the Intergovernmental Panel on Climate Change]
3. Ong HC, Mahlia TMI, Masjuki HH (2011) A review on energy scenario and sustainable energy
in Malaysia. Renew Sustain Energy Rev 15:639–647
4. Aditya L, Mahlia TMI, Rismanchi B, Ng HM, Hasan MH, Metselaar HSC, Muraza O, Aditiya
HB (2017) A review on insulation materials for energy conservation in buildings. Renew Sustain
Energy Rev 73:1352–1365
5. Huiothers SC (2000) Building energy efficiency standards in Hong Kong and mainland China.
In: Proceedings of the 2000 ACEEE summer study on energy efficiency in buildings, 2000, pp
20–25
6. Choi K, Kim JH, Shin K (2004) Economic feasibility analysis of roadway capacity expansion
with accounting traffic noise barrier cost. KSCE J Civil Eng 8:117–127
7. Mirzahosseini M, Riding KA (2015) Influence of different particle sizes on reactivity of finely
ground glass as supplementary cementitious material (SCM). Cement Concr Compos 56:95–
105
8. Lachemi M, Şahmaran M, Hossain KMA, Lotfy A, Shehata M (2010) Properties of controlled
low-strength materials incorporating cement kiln dust and slag. Cement Concr Compos
32(8):623–629
9. Sheen YN, Zhang LH, Le DH (2013) Engineering properties of soil-based controlled low-
strength materials as slag partially substitutes to Portland cement. Constr Build Mater 48:822–
829
10. Youssef AM, Assem F, Essam M, Elaaser M, Ibrahim O, Mahmoud M, Abd El- Salam M
(2019) Development of a novel bionanocomposite material and its use in packaging of Rams
cheese. Food Chem 270:467–75
11. Youssef AM, El-Sayed SM, El-Sayed HS, Salama HH, Mohamed HF, Abd El- Salam MA
(2018) Novel bionanocomposite materials used for packaging skimmed milk acid coagulated
cheese (Karish). Int J Biol Micromoles 115(2018):1002–11
12. Youssef AM, El-Nahrawy AM, Abou Hammad AB (2017) Sol-gel synthesis and characteri-
zations of hybrid chitosan-PEG/calcium silicate nanocomposite modified with ZnO-NPs and
(E102) for optical and antibacterial applications. Int J Biol Macromol 97:561–567
13. Youssef AM, El- Sayed SM (2018) Bionanocomposites materials for food packaging
applications: concepts and future Outlook. Carbohydr Polym 193:19–27
14. BumbudsanpharokeN Ko S (2015) Nano-food packaging: an overview of market, migration
research, and safety regulations. J Food Sci 80:R910
15. Youssef AM (2013) Polymer nanocomposites as a new trend for packaging applications. Polym
Plast Technol Eng 52:635–660
16. Anbazhagan M, Ramachandran S, Subramanian P, Nachimuthu R, Gothandam KM, Ranjan
S, Dasgupta N, Lichtfouse E (2016) Nanoscience in food and agriculture “nanomaterials:
classification, biological synthesis and characterization”, 3rd ed. Springer, pp 31–71
17. Youssef AM, El- Sayed SM (2018) Bionanocomposites materials for food packaging
applications: concepts and future Outlook. Carbohydr Polym 193:19–27
Chapter 12
Protection of Nine-Phase Transmission
Line Using Biorthogonal-2.2 Wavelet
Transform
Gaurav Kapoor
1 Introduction
An increase in the inevitability of electrical power has been perceived by the people
of the modern generation. The electrical power transfer potentiality of the currently
operating power transmission systems ought to be augmented in order to assist the
significant increase in the necessity of electrical energy. In the literature, NPTL’s
have been suggested as an imminent replacement of the prevalent configuration of
the electrical power transmission system which has the prospective for transferring
the large extent of electrical energy.
The feasibility of fault occurrence on the NPTL is more as compared with the
DCTL. Thus, accurate recognition of the faults in the NPTL turns out to be very
decisive for mitigating the loss of gain and providing fast renovates.
Several newly reported research works addressed the issue related to fault recog-
nition and categorization in TL’s. Some important research attempts are presented in
concise here in this section. Recently, fuzzy logic has been employed for micro-grid
protection [1]. The TPTL has been protected using WT in [2]. In [3], ConvNet has
been applied for fault recognition in micro-grid. In [4, 5], WT has been used for fault
categorization in SPTL and SCCDCTL, respectively. Mathematical morphology has
been applied for SPTL protection in [6]. ANN and phasor data have been used for
islanding recognition in the smart grid [7]. HHT has been used for SPTL protection
in [8]. In [9], POVMD and WPNRVFLN have been employed for fault recognition in
SCDCTL. Mathematical morphology and data mining-based techniques have been
used for high impedance recognition [10].
In this work, a novel tool, i.e., the biorthogonal-2.2 wavelet transform (BWT)
is used for NPTL protection. No such type of work has been reported yet to the
G. Kapoor (B)
Modi Institute of Technology, Kota, India
e-mail: gaurav.kapoor019@gmail.com

https://doi.org/10.1007/978-981-15-8820-4_12
120 G. Kapoor
Load-1
Nine Phase
765 kV source Transmission 765 kV source
Relay Line
Load-2
Bus-1 Fault Bus-2
Fig. 1 The graphic of 765 kV nine-phase power system
best of the knowledge of the author. The results show that the BWT powerfully
recognizes and categorizes the faults and the consistency of BWT is not perceptive
to the variation in various fault factors.
This article is structured as: The specifications of NPTL are reported in Sect. 2.
Section 3 shows the flow diagram of BWT. Section 4 is dedicated to the discussion
of the response of BWT. Section 5 completes the paper.
2 The Specifications of NPTL
The system of a 765 kV NPTL is designed using MATLAB. Figure 1 shows the
illustration of NPTL. The NPTL has a rating of 765 kV, 50 Hz, and has a total length
of 200 km. The NPTL is separated into two zones of length 100 km both. Two loads
of 300 MW and 150 MVAr each are connected at the receiving end of NPTL. The
combination of relay and transducers is connected near bus-1 for the relaying of the
total length of NPTL.
3 BWT-Based Protection Technique
Figure 2 shows the process of BWT with the following steps:
Step-1 Nine-phase currents are recorded through transducers installed at bus-1.

Step-2 BWT is employed to estimate the BWT outputs of phase currents.
Step-3 The phase will be declared as the faulty phase if its BWT output has a larger
amplitude as compared to the output of the healthy phase under a faulty
situation.
12 Protection of Nine-Phase Transmission Line … 121
Record nine
phase current
signals
Signal processing using BWT
Calculate BWT
coefficients
No
Is |BWT coefficient| faulted phase
> No fault
|BWT coefficient| healthy phase
Yes
Simultaneous fault detection

and
faulty phase recognition
Fig. 2 Flow chart of BWT
4 Performance Assessment
To substantiate the ability of the BWT, the simulation effort has been carried out for
numerous faults. The simulation outcomes of the study are examined in the successive
subsections.
4.1 Response of BWT for Healthy Situation
Figure 3 shows the nine-phase currents and voltages for no-fault. Figure 4 exemplifies
the output of BWT for no-fault. Table 1 reports the results of BWT for no-fault.
4.2 Response of BWT for Fault Switching Time Variation
The BWT is investigated for variation in fault switching time (FST). Figure 5 depicts
the ABCEGHI-g fault at 100 km at 0.05 s among RF = 2.15 and RG = 3.15 .
Figure 6 shows the output of BWT for the ABCEGHI-g fault simulated at 0.05 s.
The fault factors for all the fault cases are set as: T = 0.05 s, FL = 100 km, RF =
2.15 , and RG = 3.15 . Tables 2, 3, 4, 5, and 6 tabularizes the results for variation
in FST. It is inspected from Tables 2, 3, 4, 5, and 6 that the variation in FST does not
manipulate the operation of BWT.
122 G. Kapoor
Fig. 3 Nine-phase currents and voltages for no-fault
a b c
d e f
g h i
Fig. 4 BWT outputs for no-fault
Table 1 Response of BWT

BWT outputs
for no-fault
Phase-A Phase-B Phase-C
2.1264 × 103 2.4245 × 103 2.3967 × 103
Phase-D Phase-E Phase-F
2.3212 × 103 2.3690 × 103 2.3704 × 103
Phase-G Phase-H Phase-I
2.4514 × 103 2.1949 × 103 2.1032 × 103
Fig. 5 ABCEGHI-g fault at 100 km at 0.05 s among RF = 2.15 and RG = 3.15
a b c
d e f
g h i
Fig. 6 BWT outputs for ABCEGHI-g fault at 100 km at 0.05 s among RF = 2.15 and RG =
3.15
Table 2 Response of BWT for ABCEGHI-g fault at 100 km at 0.05 s among RF = 2.15 and RG
= 3.15
Fault—ABCEGHI-g (FST = 0.05 s)
Phase BWT output Phase BWT output Phase BWT output
A 6.7582 × 104 D 3.8703 × 103 G 5.4475 × 104
B 6.7227 × 104 E 3.0215 × 104 H 6.1760 × 104
C 5.7644 × 104 F 2.3550 × 103 I 5.8108 × 104
124 G. Kapoor
Table 3 Response of BWT for ABGHI-g fault at 100 km at 0.1 s among RF = 2.15 and RG =
3.15
Fault—ABGHI-g (FST = 0.1 s)
A 5.7034 × 104 D 2.1756 × 103 G 6.2186 × 104
B 5.4816 × 104 E 2.3311 × 103 H 5.4787 × 104
C 2.5569 × 103 F 2.4938 × 103 I 6.1941 × 104
Table 4 Response of BWT for BCDEF-g fault at 100 km at 0.17 s among RF = 2.15 and RG =
3.15
Fault—BCDEF-g (FST = 0.17 s)
A 2.8030 × 103 D 6.3673 × 104 G 2.3346 × 103
B 5.4067 × 104 E 6.0301 × 104 H 2.5046 × 103
C 5.2134 × 104 F 5.9313 × 104 I 2.3395 × 103
Table 5 Response of BWT for ABHI-g fault at 100 km at 0.08 s among RF = 2.15 and RG =
3.15
Fault—ABHI-g (FST = 0.08 s)
A 5.5787 × 104 D 2.3102 × 103 G 2.6451 × 103
B 4.9309 × 104 E 2.3134 × 103 H 5.3814 × 104
C 2.6365 × 103 F 2.4714 × 103 I 4.5725 × 104
Table 6 Response of BWT for DEF-g fault at 100 km at 0.2 s among RF = 2.15 and RG = 3.15

Fault—DEF-g (FST = 0.2 s)
A 2.3810 × 103 D 5.6832 × 104 G 2.3544 × 103
B 2.1445 × 103 E 5.9226 × 104 H 2.2560 × 103
C 2.1369 × 103 F 5.7041 × 104 I 2.4192 × 103
4.3 Response of BWT for Near-in Relay Faults
The efficiency of the BWT is tested for various near-in relay faults on the NPTL.
Figure 7 depicts the DEFGHI-g near-in relay fault current at 5 km at 0.12 s among
RF = 3.5 and RG = 2.5 . Figure 8 shows the BWT coefficients for the DEFGHI-g
fault simulated at 5 km. The fault factors for all the fault cases are: T = 0.12 s, RF =
3.5 and RG = 2.5 . Tables 7, 8, 9, 10, and 11 details the results of the BWT for
Fig. 7 DEFGHI-g near-in relay fault at 5 km at 0.12 s among RF = 3.5 and RG = 2.5
a b c
d e f
g h i
Fig. 8 BWT outputs for DEFGHI-g near-in fault at 5 km among RF = 3.5 and RG = 2.5
Table 7 Response of BWT for DEFGHI-g fault at 5 km

Fault—DEFGHI-g (5 km)
A 131.6663 D 2.6131 × 103 G 3.0297 × 103
B 456.0025 E 6.5031 × 103 H 3.9445 × 103
C 108.0168 F 1.8497 × 103 I 3.3792 × 103
five different near-in relay faults. It is confirmed from Tables 7, 8, 9, 10, and 11 that
the BWT has the ability to detect the near-in relay faults precisely.
126 G. Kapoor
Table 8 Response of BWT for ABCEF-g fault at 6 km

Fault—ABCEF-g (6 km)
A 3.0852 × 103 D 818.7633 G 26.7045
B 9.0172 × 103 E 6.2284 × 103 H 194.3373
C 2.6214 × 103 F 2.5241 × 103 I 93.5313
Table 9 Response of BWT for GHI-g fault at 7 km

Fault—GHI-g (7 km)
A 86.5409 D 30.4280 G 5.1182 × 103
B 299.8279 E 394.6009 H 5.2485 × 103
C 89.1173 F 87.1063 I 3.5345 × 103
Table 10 Response of BWT for ABC-g fault at 8 km

Fault—ABC-g (8 km)
A 2.9549 × 103 D 61.3253 G 65.5750
B 1.0461 × 103 E 272.2014 H 249.4814
C 2.9781 × 103 F 81.4085 I 104.3835
Table 11 Response of BWT for DEF-g fault at 9 km

Fault—DEF-g (9 km)
A 106.9428 D 5.0679 × 103 G 94.3637
B 312.8468 E 8.8515 × 103 H 389.2334
C 173.8100 F 3.4420 × 103 I 103.6978
4.4 Response of BWT for Far-End Relay Faults
The BWT has been explored for different far-end relay faults. Figure 9 illustrates the
ABCDEFG-g far-end relay fault at 195 km at 0.06 s among RF = 4.5 and RG =
3.5 . Figure 10 shows the BWT coefficients for the ABCDEFG-g fault simulated
at 195 km. The fault factors chosen for all the fault cases are: T = 0.06 s, RF = 4.5
and RG = 3.5 . Tables 12, 13, 14, 15, and 16 report the results for various far-end
relay faults. It is inspected from Tables 12, 13, 14, 15, and 16 that the effectiveness
of BWT remains impassive for different far-end relay faults.
Fig. 9 ABCDEFG-g fault at 195 km at 0.06 s among RF = 4.5 and RG = 3.5
a b c
d e f
g h i
Fig. 10 BWT outputs for ABCDEFG-g fault at 195 km at 0.06 s among RF = 4.5 and RG = 3.5
Table 12 Response of BWT for ABCDEFG-g fault at 195 km

Fault—ABCDEFG-g (195 km)
A 6.1249 × 104 D 6.6442 × 104 G 3.4487 × 104
B 5.9615 × 104 E 6.1556 × 104 H 2.1853 × 103
C 5.3307 × 104 F 5.7471 × 104 I 4.2411 × 103
4.5 Response of BWT for Variation in Fault Resistance
The BWT is investigated for variation in fault resistances. Figure 11 depicts the
ABCDEFHI-g fault at 100 km at 0.07 s among RF = 10 and RG = 0.001 .
128 G. Kapoor
Table 13 Response of BWT for ABDEF-g fault at 196 km

Fault—ABDEF-g (196 km)
A 5.3989 × 104 D 5.7792 × 104 G 2.4971 × 103
B 4.7741 × 104 E 5.9487 × 104 H 2.3224 × 103
C 2.8607 × 103 F 5.9446 × 104 I 2.4280 × 103
Table 14 Response of BWT for EFGHI-g fault at 197 km

Fault—EFGHI-g (197 km)
A 2.3752 × 103 D 2.7792 × 103 G 6.1323 × 104
B 2.3674 × 103 E 5.0953 × 104 H 5.4694 × 104
C 2.1085 × 103 F 4.8050 × 104 I 5.5860 × 104
Table 15 Response of BWT for ABDE-g fault at 198 km

Fault—ABDE-g (198 km)
A 5.6799 × 104 D 5.5657 × 104 G 2.2046 × 103
B 5.1894 × 104 E 4.9065 × 104 H 2.4423 × 103
C 2.5343 × 103 F 2.5271 × 103 I 2.3698 × 103
Table 16 Response of BWT for ABC-g fault at 199 km

Fault—ABC-g (199 km)
A 5.9040 × 104 D 2.1843 × 103 G 2.1728 × 103
B 5.5764 × 104 E 2.3957 × 103 H 2.2799 × 103
C 5.1070 × 104 F 2.2733 × 103 I 2.1238 × 103
Fig. 11 ABCDEFHI-g fault at 0.07 s at 100 km among RF = 10 and RG = 0.001

Figure 12 shows the BWT coefficients for the ABCDEFHI-g fault simulated among
RF = 10 . The fault factors for all the fault cases are set as T = 0.07 s, FL = 100 km,
and RG = 0.001 . Tables 17, 18, 19, 20, and 21 tabularizes the results for variation
in fault resistances. It is inspected from Tables 17, 18, 19, 20, and 21 that variation
in the fault resistances does not manipulate the working of the BWT.
a b c
d e f
g h i
Fig. 12 BWT outputs for ABCDEFHI-g fault at 0.07 s at 100 km among RF = 10 and RG =
0.001
Table 17 Response of BWT for ABCDEFHI-g fault among RF = 10

Fault—ABCDEFHI-g (10 )
A 6.8829 × 104 D 6.5690 × 104 G 3.2895 × 103
B 6.5526 × 104 E 6.8908 × 104 H 6.4465 × 104
C 6.7075 × 104 F 6.7971 × 104 I 5.9149 × 104
Table 18 Response of BWT for BCDEF-g fault among RF = 40

Fault—BCDEF-g (40 )
A 2.7621 × 103 D 3.8084 × 104 G 2.2360 × 103
B 3.2730 × 104 E 3.6160 × 104 H 2.3975 × 103
C 3.3935 × 104 F 3.9879 × 104 I 2.3335 × 103
130 G. Kapoor
Table 19 Response of BWT for ACGH-g fault among RF = 70

Fault—ACGH-g (70 )
A 2.6259 × 104 D 2.4778 × 103 G 2.4334 × 104
B 3.0921 × 103 E 2.5364 × 103 H 2.8309 × 104
C 2.5700 × 104 F 2.2849 × 103 I 2.6976 × 103
Table 20 Response of BWT for DEF-g fault among RF = 100

Fault—DEF-g (100 )
A 2.2718 × 103 D 1.9765 × 104 G 2.3222 × 103
B 2.2663 × 103 E 1.8950 × 104 H 2.3157 × 103
C 2.3288 × 103 F 1.9448 × 104 I 2.1198 × 103
Table 21 Response of BWT for AGHI-g fault among RF = 150

Fault—AGHI-g (150 )
A 1.7154 × 104 D 2.4446 × 103 G 2.0733 × 104
B 2.2694 × 103 E 2.2100 × 103 H 1.9942 × 104
C 3.1200 × 103 F 2.4650 × 103 I 2.1785 × 104
5 Conclusion
This work showed an improved revelation using BWT to recognize and categorize
the faults occurring on NPTL. The BWT coefficients of the nine-phase currents of the
NPTL which are measured at one-end are employed by BWT for fault recognition,
categorization, and faulty phase identification. The value of fault resistance is varied
from 10 to 150 , the position of fault for the near-in relay faults is varied from
5 to 9 km, and the position of fault for the far-end relay faults is varied from 195
to 199 km. The simulation studies support the consistency of BWT under extensive
variations in fault type, location, resistance, and switching time. From the results, it is
noticeable that there is an evident intolerance between the fault and no-fault situations
and establishes the potential of the BWT-based fault recognition and faulty phase
categorization technique by recognizing the faults correctly.
References
1. Chaitanya BK, Soni AK, Yadav A (2018) Communication assisted fuzzy based adaptive
protective relaying scheme for microgrid. J Power Technol 98(1):57–69
2. Kapoor G (2018) Wavelet transform based detection and classification of multi-location three
phase to ground faults in twelve phase transmission line. Majlesi J Mechatron Systems 7(4):47–
60
3. Manohar M, Koley E, Ghosh S (2019) Enhancing resilience of PV-fed microgrid by improved
relaying and differentiating between inverter faults and distribution line faults. Electric Power
Energy Syst (Elsevier) 108:271–279
4. Kapoor G (2018) Six phase transmission line boundary protection using wavelet transform.
In: Proceedings of the 8th IEEE India international conference on power electronics (IICPE).
IEEE. Jaipur, India (2018)
5. Gautam N, Ali S, Kapoor G (2018) Detection of fault in series capacitor compensated double
circuit transmission line using wavelet transform. In: Proceedings of the IEEE international
conference on computing, power and communication technologies (GUCON), pp 769–773.
IEEE. Greater Noida, India (2018)
6. Kapoor G (2018) Six phase transmission line boundary protection using mathematical
morphology. In: Proceedings of the IEEE international conference on computing, power and
communication technologies (GUCON), pp 857–861. IEEE. Greater Noida, India (2018)
7. Kumar D, Bhowmik PS (2018) Artificial neural network and phasor data-based islanding
detection in smart grid. IET Gener Transm Distrib 12(21):5843–5850
8. Kapoor G (2019) Detection and classification of single line to ground boundary faults in a
138 kV six phase transmission line using Hilbert Huang transform. i-manager’s J Electric Eng
12(3):28–41 (2019)
9. Sahani M, Dash P K (2019) Fault location estimation for series-compensated double-circuit
transmission line using parameter optimized variational mode decomposition and weighted
P-norm random vector functional link network. Appl Soft Comput J (Elsevier), 1–18 (2019)
10. Sekar K, Mohanty NK (2018) Data mining-based high impedance fault detection using
mathematical morphology. Comput Electric Eng (Elsevier) 69:129–141
Chapter 13
An Intelligent Hybrid Model
for Forecasting of Heart and Diabetes
Diseases with SMO and ANN
Shalini, Pawan Kumar Saini, and Yatendra Mohan Sharma
1 Introduction
Nowadays, among several human life killer diseases, heart and diabetes are the two
most widely recognized causes. Related literature of medical science illustrates that
only forecasting such diseases at the initial period is only a single hope for saving
a life. However, verification of diseases with its causes is an effort of the physician
but a huge amount of common and related symptoms make the decision step harder.
Additionally, incorrect or deferred forecast information may cause of threat or more
harm to a patient. However, to boost disease recognition process and treat patients
untimely huge medical experts consume different practices but due to an intricate
system of the heart and huge common symptoms of diabetic disease active schemes
have its fashionable restraint. Therefore, the field is still open for modernization. To
improve disease recognition quality with elevated accuracy a naïve hybrid forecasting
model based on neural network and SMO theorem is employed in this paper. The
experimental fallouts have shown the importance of the proposed method in the
classification of heart and diabetic diseases.
Shalini
Arya Institute of Engineering Technology & Management, Jaipur, Rajasthan, India
P. K. Saini
Modern Institute of Technology & Research Centre, Alwar, Rajasthan, India
Y. M. Sharma (B)
Biff and Bright College of Engineering & Technology, Dudu, Jaipur, Rajasthan, India
e-mail: er.shalini.pathak7@gmail.com; alka.choubey2@gmail.com

https://doi.org/10.1007/978-981-15-8820-4_13
134 Shalini et al.
2 Related Efforts
Among several obtainable techniques, a good amount of investigators has demon-

strated the rewards of ANN and SMO theorem in the skillful classification of heart
and diabetes diseases, some modern study has discussed in [1–7]. To improve the
forecasting accuracy of heart and diabetes diseases a naïve hybrid model that consists
of combine features of ANN and FNN in sole form has illustrated in [8]. For demon-
strating the effectiveness of the intended approach, the proposed methods have evalu-
ated with two benchmarked dataset, Pima Indians diabetes and Heart disease dataset.
The model exploits k-fold cross-validation scheme to crack the selected dataset into a
k part and each time one part is consumed for training and rest remains has utilized for
testing purposes. The proportional fallouts demonstrate that the presented approach
attain high accuracy level when match up to the records at UCI and related studies for
heart disease. With the integration of NN and GA (genetic algorithm) into a single
model, a naïve approach has made an endeavor to improve forecasting accuracy of
heart diseases [9]. GA-based procedure utilized for determined an appropriate prelim-
inary weight of neural network. To show the effectiveness of presented method, the
investigators have evaluated an act of proposed scheme with five different bench-
marked dataset with utilizing the functionality of 10-fold cross-validation scheme.
Experimental upshots depicted the efficiency of presented scheme. With the compar-
ative estimation of SVM, DT, NB, KNN, and ANN methods for forecasting heart
diseases a naïve investigation demonstrates that NB and SVM are the most suitable
methods for recognizing heart diseases in a speedy form [10]. A naïve model with
MLPNN (Multilayer Perceptron Neural Network) and BP (Back Propagation) has
utilized to improve the attained accuracy level of heart diseases [11]. To demonstrate
the uses of presented approach it has evaluated with the 303 patients’ records and
results prove the efficiency of published approach.
Additionally, the above-discussed methods, a number of algorithms, have been
presented by different investigators [12–16] for forecasting of heart and diabetes
diseases but elaboration of each and every approach is not possible to comprise in
this paper. On the other hand, much of published work has demonstrated that ANN-
based approaches are significantly well performed and can be utilized for prediction
and classification purposes but in front of ensemble classification-based approaches
the simple ANN-based algorithm has shown stumpy performance, hence there is
appalling need of optimization.
3 Proposed Approach
Motivated from related literature that in place of any sole process, the hybrid methods
are most suitable for forecasting of diseases and utilization of only the most relative
attributes from an entire attribute list of opted dataset is a worthy decision that
significantly improves an act of build procedures, the anticipated approach mix the
13 An Intelligent Hybrid Model for Forecasting … 135
muscle of feed-forward neural network MLP (Multilayer Perceptron) and SMO into
a single form with the exploitation of correlation-based feature selection and breadth-
first search mechanism to discover and integrate only most relative features of opted
medical dataset for the forecasting process of diseases. Typically, MLP is a form
of ANN in which build network is connected with single or added hidden layers
from input to output layer. At initial phase, proposed model preprocessed the opted
dataset, remove incomplete data entries, and discover the most optimal attributes
set for forecasting process of diseases. Subsequent to preprocessing process, the
build model breaks the dataset into training and testing phase and starts the disease
forecasting process with SMO approach. The approach trains the build model with
a part of opted dataset and performs a testing phase with untouched part, utilizes
10-fold cross-validation process in which build model utilizes one part for training
and other exploit for the testing phase. Once the SMO approach completes its task
the build mechanism delivers the misclassified data to MLP method that significantly
improves the forecasting accuracy of diseases. Additionally, the MLP technique uses
3 hidden layer to enhance an amount of accurate forecasting, following Fig. 1 shows
the structure of MLP method with selected optimal features set for forecasting of
diabetes, among total 9 features the proposed approach only utilizes 4 features as
shown in Fig. 1.
Where Ds stand for opted dataset of diseases, collected from benchmarked collec-
tion library UCI. PDs are preprocessed data set, after removing incomplete data and
selection of an optimal feature set (OFs). TRDs and TDs denote the training and
testing part of preprocessed dataset. MD is misclassified data (Fig. 2).
Fig. 1 Utilized multilayer perceptron model for forecasting of diabetes diseases

136 Shalini et al.
Fig. 2 Proposed methodology for diseases forecasting process
4 Result Analysis
To analyze an act of the build method two different medical dataset heart and diabetes
taken into account have been collected from well-accepted library UCI. Fair evalua-
tion simulation has been done with an identical Pima Indian diabetes data set that has
been considered in [2, 7], enclose 768 diabetes patients record with total of 9 features.
However, in place of 9 features, the build approach utilizes only 4 features for the
forecasting process of diabetes. The following figure illustrates the significance and
acceptability of the build method (Fig. 3).
The presented values in the above figure illustrate that proposed approach has
attained high accuracy in comparison to other methods, presented in [2, 7]. However,
difference between DIM and the presented method of this paper is not huge but in
medical area a little amount has its significance therefore the proposed approach
is most suitable in place of other obtainable methods. To discover an efficiency of
proposed method it has again evaluated with a different disease dataset that has to
consider by [4, 17–19], heart diseases Cleveland dataset that holds down reports of
303 patients with 14 attributes. However, as alike previous experiment, the build
model opted only most relative attribute set among the list of attributes, elect only
7. The experimental result depicted in the figure has shown that the build method
has attained the finest accuracy 97.02% in comparison to other recent proposed
120 98 98.44
89.56
Accuracy(%)
100 73.4 80.21

80
60
40
20
0
PNN ANN with GA GRNN Disease Proposed
Influence Approach
Measure
(DIM)
Diseases Forecasting Methods
Fig. 3 Comparative Analysis over diabetes diseases dataset

13 An Intelligent Hybrid Model for Forecasting … 137
98 97.02
96
Accuracy(%)
94.01
94 93
92 91.1
90
88
ANN-Fuzzy-AHP NB+KNN [18] HECFNN [19] Proposed Approach
[17]
Forecasting Procedures
Fig. 4 Comparative analysis over heart diseases dataset
approaches. The both of above comparative fallouts of the build method demonstrate
its efficiency over the other obtainable methods.
As the experimental result depicted in the Fig. 4, the build method has attained
finest accuracy 97.02% in comparison to other recent proposed approaches. The both
of the above comparative fallouts of the build method demonstrate its efficiency over
the other obtainable methods.
5 Conclusion
Over past decades, heart and diabetes diseases are the two main causes of early
mortality of human life. However, many investigators exploit different methodologies
to aid medical professionals in the direction to save life but each and every obtainable
method struggle due to the intricacy structure of the heart and human body. Recently
machine learning-based method attained huge fame in medical field to aid related
field professionals with quick responses to make life-saving decisions. This paper
intends a naïve hybrid forecasting procedure of heart and diabetes diseases on the base
of SMO and ANN methods. The proposed approach significantly improves the fore-
casting accuracy of diabetes and heart diseases with attaining 98.44% and 97.02%.
Experimental evaluation demonstrates that the proposed approach has attained high
accuracy in comparison to other modern algorithms and is more suitable for fore-
casting procedure. Future work will examine an act of the build method with other
dataset and with inclusion of different feature selection procedures to enhance an
accuracy percentage.
References
1. Rahiml P, Jafarian A (2016) Prediction of diabetes by using artificial neural network, logistic
regression statistical model and combination of them. Bulletin de la Société Royale des Sciences
de Liège 85:1148–1164
2. Sujarani P, Kalaiselvi K (2018) Prediction of diabetes using artificial neural networks: a review.
J Adv Res Dynam Control Syst 10(4):67–72
138 Shalini et al.
3. Dbritto R, Srinivasaraghavan A, Joseph V (2016) Comparative analysis of accuracy on heart

disease prediction using classification methods. Int J Appl Inform Syst 2249–0868 (2016)
4. Nagamani T, Logeswari S, Gomathy B (2019) Heart disease prediction using data mining with
mapreduce algorithm. Int J Innov Technol Explor Eng (IJITEE) 8(3), 137–140
5. Srivastava S, Sharma L, Sharma V, Kumar A, Darbari H (2019) Prediction of diabetes using
artificial neural network approach. In: ICoEVCI. Lecture Notes in Electrical Engineering, pp
679–687
6. Sharma P, Saxena S, Sharma YM (2018) An efficient decision support model based on ensemble
framework of data mining features assortment & classification process. In: Proceedings of the
international conference on communication and electronics systems (ICCES 2018). IEEE, pp
487–491
7. Baiju BV, John Aravindhar D (2019) Disease influence measure based diabetic prediction
with medical data set using data mining. In: 1st International conference on innovations in
information and communication technology (ICIICT). IEEE, pp 1–6
8. Kahramanli H, Allahverdi N (2008) Design of a hybrid system for the diabetes and heart
diseases. Expert Syst Appl, Elsevier 35:82–89
9. Haq AU, Li JP, Memon MH, Nazir S, Sun R (2018) A hybrid intelligent system framework for
the prediction of heart disease using machine learning algorithms. Hindawi, Mobile Information
Systems, Article ID 3860146, pp 1–22
10. Beyene C, Kamat P (2018) Survey on prediction and analysis the occurrence of heart disease
using data mining techniques. Int J Pure Appl Math 118:165–174
11. Singh P, Singh S, Pandi-Jain GS (2018) Effective heart disease prediction system using data
mining techniques. Int J Nanomed:121–124
12. Liu X, Wang X, Su Q, Zhang M, Zhu Y, Wang Q, Wang Q (2017) A hybrid classification
system for heart disease diagnosis based on the RFRS method. Hindawi Comput Math Method
Med:1–11
13. Priscila SS, Hemalatha M (2017) Diagnosis of heart disease with particle bee-neural network.
Special section: computational life sciences and smarter technological advancement. Biomed
Res India, pp 1–7(2017)
14. Tayefia M, Tajfard M, Saffar S, Hanachi P, Amirabadizadeh AR, Esmaeily H, Taghipour A,
Ferns GA, Moohebati M, Ghayour-Mobarhan M (2017) hs-CRP is strongly associated with
coronary heart disease (CHD): a data mining approach using decision tree algorithm. Comput
Methods Program Biomed, ELSEVIER, 141(2017):105–109
15. Wijaya SH, Pamungkas GT, Sulthan MB (2018) Improving classifier performance using particle
swarm optimization on heart disease detection. In: IEEE international seminar on application
for technology of information and communication (iSemantic), pp 603–608
16. Poornima V, Gladis D (2018) A novel approach for diagnosing heart disease with hybrid
classifier. Biomed Res 29(11): 2274–2280
17. Samuel OW et al (2017) An integrated decision support system based on ANN and Fuzzy_AHP
for heart failure risk prediction. Expert Syst Appl 68:163–172
18. Malav A, Kadam K, Kamat P (2017) Prediction Of heart disease using k-means and artificial
neural network as hybrid approach to improve accuracy. Int J Eng Technol 9(4):3081–3085
19. Alkhasawneh MSh (2019) Hybrid cascade forward neural network with elman neural network
for disease prediction. Arab J Sci Eng, Springer:1–12
Chapter 14
Power Quality Assessment of Solar PV
Standalone System Using Various
DC-DC Converters
Surbhi Shringi, Santosh Kumar Sharma, and Utkarsh Gupta
1 Introduction
Today the increasing energy demands with urbanization, industrialization, and

increasing population have led to the need to find substitutes for conventional energy
sources. Many renewable sources have lead to replacing conventional sources with
time [1, 2].
The most used types of renewable power sources include solar and wind. Both
of these sources use various types of power electronics devices in order to get
synchronized with the grid [3].
The main issues that are created with the power electronics devices are the ripples
and the Total Harmonic distortion which are introduced in the power supply. Hence
in order to reduce these above-discussed problems, THD analysis of various DC-
DC converters is done in order to find the suitable and best converter amongst the
three [4, 5].
S. Shringi (B) · S. K. Sharma · U. Gupta

Rajasthan Technical University, Kota, Rajasthan, India
e-mail: surbhishringi96@gmail.com
S. K. Sharma
e-mail: santoshelect@yahoo.com
U. Gupta
e-mail: utkarshgupta3095@gmail.com

https://doi.org/10.1007/978-981-15-8820-4_14
140 S. Shringi et al.
2 System Description with Simulation Models and Results
The system consists of a PV array which acts as a power source keeping the irradiation
and the temperature to be fixed which is further given to the DC-DC converter
followed by the P&O MPPT technique to extract maximum power from the PV
array this power is collected and measured at the DC link as it is considered as one of
the parameters for power quality assessment. This DC voltage is converted into AC
using the Voltage source converter and then this is connected with the Power grid or
with the Load (Fig. 1).
2.1 System with Boost Converter
The simulation model of the system with Boost converter is shown in Fig. 2.
Figure 3 shows the output voltage at DC link of the Boost converter.
Measurements
DC-DC
VSC LC-Filter LOAD
CONVERTER
PV-Array
Fig. 1 Block diagram of the system
Fig. 2 Simulation diagram of the system using Boost converter

14 Power Quality Assessment of Solar PV … 141
Input Voltage 400
200
0
0 1 2 3 4
1500
Output Voltage
1000
500
0
0 1 2 3 4
Time
Fig. 3 Output Voltage of the DC link with Boost converter
The THD analysis of the VSC output waveforms is done whose results are shown
below in Fig. 4.
Fig. 4 THD analysis of VSC waveforms with Boost converter

2.2 System with Buck Converter
The simulation model of the system using Buck converter is shown in Fig. 5.
Figure 6 shows the output voltage at the DC link with the Buck converter.
below in Fig. 7.
Fig. 5 Simulation model of the system with Buck converter
400
Input Voltage
200
0
0 0.5 1 1.5 2 2.5 3 3.5 4
400
Output Voltage
200
0
0 0.5 1 1.5 2 2.5 3 3.5 4
Time
Fig. 6 Output voltage of the DC link with Buck converter

Signal
Selected signal: 200 cycles. FFT window (in red): 1 cycles
0.5
Signal mag.
-0.5
3.55 3.6 3.65 3.7 3.75 3.8 3.85 3.9 3.95 4
FFT analysis
Fundamental (50Hz) = 0.2615 , THD= 37.32%
30
Mag (% of Fundamental)
20
10
0
0 100 200 300 400 500 600 700 800 900 1000
Fig. 7 THD analysis of VSC waveforms with Buck converter
2.3 System with Buck-Boost Converter
The simulation model of the system using Buck-Boost converter is shown in Fig. 8.
Figure 9 shows the output voltage at the DC link with the Buck-Boost converter.
Fig. 8 Simulation model of the system with Buck-Boost converter

Input Voltage 300
200
100
0
0 0.5 1 1.5 2 2.5 3 3.5 4
1000
Output Voltage
500
0
0 0.5 1 1.5 2 2.5 3 3.5 4
Fig. 9 Output voltage of the DC link with Buck-Boost converter
in Fig. 10.
Fig. 10 THD analysis of VSC waveforms with Buck-Boost converter

2.4 System with Cuk Converter
The simulation model of the system using Cuk converter is shown in Fig. 11.
Figure 12 shows the output voltage at the DC link with the Cuk converter.
in Fig. 13.
Fig. 11 Simulation model of the system with Cuk converter
400
Input Voltage
300
200
100
0
0 0.5 1 1.5 2 2.5 3 3.5 4
1000
Output voltage
500
0
0 0.5 1 1.5 2 2.5 3 3.5 4
Fig. 12 Output voltage of the DC link with Cuk converter

Fig. 13 THD analysis of VSC waveforms with Cuk converter
2.5 System with Zeta Converter
The simulation model of the system using Zeta converter is shown in Fig. 14.
Figure 15 shows the output voltage at the DC link with the Zeta converter.
in Fig. 16.
3 Conclusion
The analysis of various converters is done on the basis of the THD and the output
voltage response. The Zeta and Boost converters are found to be better than the other
converters while talking in terms of the THD and output voltage. The selection of
the converters could be done according to the application and the user’s needs.
Fig. 14 Simulation model of the system with Zeta converter
400
Input Voltage
300
200
100
0
0 0.5 1 1.5 2 2.5 3 3.5 4
1500
Output Voltage
1000
500
0
0 0.5 1 1.5 2 2.5 3 3.5 4
Time
Fig. 15 Output voltage of the DC link with Zeta converter

Fig. 16 THD analysis of VSC waveforms with Zeta converter
References
1. Bhattacharyya S, Samanta S (2018) DC link voltage control based power management scheme
for standalone PV systems. In: 2018 IEEE international conference on power electronics, drives
and energy systems (PEDES). https://doi.org/10.1109/pedes.2018.8707904
2. Fu J, Zhang B, Qiu D, Xiao W (2014) A novel single-switch cascaded DC-DC converter of Boost
and Buck-boost converters. In: 16th European conference on power electronics and applications.
https://doi.org/10.1109/epe.2014.6910723
3. Soheli SN, Sarowar G, Hoque MA, Hasan MS (2018) Design and analysis of a DC -DC buck
boost converter to achieve high efficiency and low voltage gain by using buck boost topology into
buck topology. In: 2018 International conference on advancement in electrical and electronic
engineering (ICAEEE). https://doi.org/10.1109/icaeee.2018.8643001
4. Plotnikov I, Braslavsky I, Ishmatov Z (2016) The mathematical simulation of DC-DC coverter
in the frequency-controlled electric drive with ultracapacitors. In: 2016 International symposium
on power electronics, electrical drives, automation and motion (SPEEDAM). https://doi.org/10.
1109/speedam.2016.7525910
5. Maroti PK, Padmanaban S, Wheeler P, Blaabjerg F, Rivera M (2017) Modified high voltage
conversion inverting cuk DC-DC converter for renewable energy application. In: IEEE southern
power electronics conference (SPEC). https://doi.org/10.1109/spec.2017.8333675
Chapter 15
Coordinated Control of UPFC-Based
Damping Controller with PID for Power
System
Amit Singhal and Ankit Tandon
1 Introduction
Flexible AC Transmission System Technology introduced in 1988 by Hingorani is an

enabling technology and provides added flexibility and can enable a line to transfer
power to the thermal rating. Unified Power Flow Controller is the best FACTS device
which can control the various power system parameters like terminal voltage, phase
angle, line impedance, etc. Therefore, it can be used not only for power flow but
also for the power stabilizing control. The unified power flow controller (UPFC) is
one of the most commonly used FACTS devices that provides the most important
performance in damping small frequency oscillation in the power system [1, 2].
A comprehensive and analytical idea for modeling of UPFC for linearised and
steady-state dynamic stability has been proposed. However, in some operating condi-
tions, the PSS may fail to stabilize the power system, especially in low-frequency
oscillations [3, 4].
2 SMIB System Without UPFC Considering Fault
A SMIB system installed without UPFC is considered (Fig. 1).
δ̇ = ω0 ω

• k1 k2 D
ω = − δ + − E q + − ω
M M M
A. Singhal (B) · A. Tandon

Jodhpur Group of Institution, Jodhpur, Rajasthan, India
e-mail: amitsinghal.2891@gmail.com

https://doi.org/10.1007/978-981-15-8820-4_15
150 A. Singhal and A. Tandon
Vt ItL ILB VB
VL
XtL XLB
Alternator
Fig. 1 A single machine infinite bus system (SMIB)
Fig. 2 Simulation model of SMIB system without UPFC

k3 k4 1
Ė
q = − δ + − E q + E f d
Td0 Td0 Td0

• k5 ka ka k6 ka 1 ka

E = − δ + ω + − E q + − E f d + Vr e f
fd Ta Ta Ta Ta Ta
With the help of these linearised equations of SMIB, we obtained a simulation

model of SMIB without UPFC considering fault (Fig. 2).
The output of this simulation is taken as angle deviation, electrical terminal voltage
and power. Simulation process is being done for duration of 30 s (Figs. 3 and 4).
3 SMIB System with UPFC
A SMIB system installed with UPFC is shown in Fig. 5. Shunt converter is connected
in parallel with the power system through an exciting transformer(ET), and second
converter is attached in series with the system through a boosting transformer(BT).
Both transformers are connected via a dc link.
The diagram of damping controller UPFC based is shown in Fig. 6.
By nonlinear model has been linearized, a dynamic linear model is obtained.
15 Coordinated Control of UPFC Based Damping Controller … 151
10000
Variation in delta 8000
6000
4000
2000
0
-2000
0 5 10 15 20 25 30
Time
1.5
1
Variation in vt
0.5
0
-0.5
-1
-1.5
0 5 10 15 20 25 30
Time
1.3
1.2
Variation in vt
1.1
1
0.9
0.8
0.7
0.6
0.5
0 5 10 15 20 25 30
Time
Fig. 3 Simulation results of SMIB system in variation of load angle, electrical power, terminal
voltage at 80% loading considering fault
4
2 x 10
Variation in Delta
1.8
1.6
1.4
1.2
1
0.8
0.6
0.4
0.2
0
0 5 10 15 20 25 30
Time
1.5
1
Variation in Pe
0.5
0
-0.5
-1
-1.5
-2
0 5 10 15 20 25 30
Time
1.3
1.2
Variation in Vt
1.1
1
0.9
0.8
0.7
0.6
0.5
0 5 10 15 20 25 30
Time
Fig. 4 Simulation results of SMIB system in variation of load angle, electrical power, terminal
voltage at 100% loading considering fault
VBt
Vt I te IB X BV
X te VEt Vb
BT
VSC-E VSC-B
IE Cdc
ET
mE ∂ ∂B
E mB
Fig. 5 SMIB system with UPFC
∆w ∆u
Kdc STW/1+STW 1+ST1/1+ST2
Gain Signal Washout Phase Compensator
Fig. 6 Structure of UPFC-based damping controller
δ̇ = w0 w
−Pe − Dw
ẇ =
M

E = −E q + E f d /Td0
E f d Ka
Ė f d = − + Vr e f − Vt
Ta Ta

E q = K 4 δ + K 3 E q + K vd Vdc + K qe m E +
K qδe δ E + K qb m B + K qδb δ B

Vt = K 5 δ + K 6 E q + K vd Vdc + K ve m E +
K vδe δ E + K vb m B + K vδb δ B

.
Vdc = K 7 δ + K 8 E q − K 9 Vdc + K ce m E +
K cδe δ E + K cb m B + K cδb δ B
With the help of these linearised equations of SMIB and UPFC, we obtained a
simulation model of SMIB with UPFC considering fault (Figs. 7, 8 and 9).
Fig. 7 Simulation diagram of SMIB with UPFC
1.6
1.4
Variation in Delta
1.2
1
0.8
0.6
0.4
0.2
0 5 10 20 25 30
Time
1.4
1.2
Variation in Pe
1
0.8
0.6
0.4
0.2
0
0 5 10 15 20 25 30
Time
1.05
1
Variation in Vt
0.95
0.9
0.85
0.8
0.75
0.7
0 5 10 15 20 25 30
Time
Fig. 8 Simulation results of SMIB system with UPFC in variation of load angle, electrical power,
terminal voltage at 80% loading
Variation in delta 2.2

2
1.8
1.6
1.4
1.2
1
0.8
0.6
0.4
0.2
0 5 10 15 20 25 30
Time
1.5
Variation in Pe
0.5
0
0 5 10 15 20 25 30
Time
1.1
1.05
Variation inVt
1
0.95
0.9
0.85
0.8
0.75
0.7
0 5 10 15 20 25 30
Time
Fig. 9 Simulation results of SMIB system with UPFC in variation of load angle, electrical power,
terminal voltage at 100% loading
The output of this simulation is taken as angle deviation, electrical power and
terminal voltage. Simulation process is being done for duration of 30 s.
4 SMIB System with Proposed PID Controller
As we can see that using UPFC, the oscillation of SMIB is reduced. But still the
system having oscillations which should be damped. So we install a PID controller
on UPFC. The input given to PID controller is dw, and output is formed from PID
controller is given as input to UPFC (Figs. 10, 11, 12 and 13).
The simulation diagram of SMIB with proposed UPFC and PID controller as
shown in figure.
du
Uref + KA/(1+s*TA)
-
∆w Kp+KI/s+Kd*s
Fig. 10 Diagram of damping controller UPFC based with PID

Fig. 11 Simulation model of SMIB with proposed UPFC with PID controller
1.25
Variation in delta
1.2
1.15
1.1
1.05
1
0.95
0.9
0.85
0 5 10 15 20 25 30
Time
3
Variation in Pe
2.5
2
1.5
1
0.5
0
0 5 10 15 20 25 30
Time
3
Variation in Pe
2.5
2
1.5
1
0.5
0
0 5 10 15 20 25 30
Time
Fig. 12 Simulation results of SMIB system with proposed UPFC with PID controller in variation
of load angle, electrical power, terminal voltage at 80% loading
The output of this simulation is taken as angle deviation, electrical terminal voltage
and power. Simulation process is being done for duration of 30 s.
Variation in delta 1.25

1.2
1.15
1.1
1.05
1
0.95
0.9
0.85
0 5 10 15 20 25 30
Time
3
Variation in Pe
2.5
2
1.5
1
0.5
0
0 5 10 15 20 25 30
Time
2.2
Variation in Vt
2
1.8
1.6
1.4
1.2
1
0.8
0 5 10 15 20 25 30
Time
Fig. 13 Simulation results of SMIB system with proposed UPFC with PID controller in variation
of load angle, electrical power, terminal voltage at 100% loading
5 Conclusion
Aim of this work is to damp the oscillation of the power system by different
controllers. In this paper, SIMULINK model of a SMIB system with a damping
controller with UPFC based is presented. With the help of this controller oscillations
of power system will be damped out. But still system has the oscillation. Then, we
will incorporate PID controller with UPFC-based damping controller. The simulink
results show that PID controller with UPFC has better execution for damped out the
local oscillations in power system.
Appendix
The value of parameter used in simulation are

Generator:
M = 2H = 8.0, D = 0, Tdo’ = 5.044, Xd = 1.0, Xq = 0.6, Xd’ = 0.3
Excitation System:
KA = 100, TA = 0.01
Transformer:
XtE = 0.1 p.u., XE = 0.1 p.u., XB = 0.1 p.u.
Transmission Line:
XBv = 0.3, Xe = 0.5

Operating conditions:
Pe = 0.8, Vt = 1.0 p.u.
Vb = 1.0 p.u.
UPFC parameter:
mE = 0.4013, mB = 0.0789
δE = −85.34780, δB = −78.21740 .
References
1. Larsen EV, Sanchez JJ, Chow JH (2005) Concept for design of FACTS controllers to damped
power swings. IEEE Trans PWRS 10(2):948–955
2. Noroozian M, Angquist L, Ghandari M, Anderson G (2008) Use of UPFC for optimal power
flow control. IEEE Trans Power Syst 12(4):1629–1634
3. Smith KS, Ran L, Penman J (2012) Dynamic modeling of a unified power flow controller. IEE
Proc Gener Transm Distrib 144(1):7
4. Wang HF (2004) Damping function of unified power flow controller. IEE Proc 146(1):81
Chapter 16
Computational Neuroscience and Its
Applications: A Review
Aisha Jangid, Laxmi Chaudhary, and Komal Sharma
1 Introduction
1.1 Central Nervous System (CNS):The Human Brain
Central nervous system (CNS): The central nervous system (CNS) is your body’s
centre of decision and communication. It consists of the brain and the spinal cord, and
the peripheral nervous system (PNS) is made of nerves. They control and regulate
every part of daily life from breathing and blinking to help you memorize facts for a
test. The human brain is the command centre for the human nervous system. Nerves
reach from your brain to your face, ears, eyes, nose and spinal cord and from the
spinal cord to the rest of your body [1]. Sensory nerves collect the information from
the environment which sends it to spinal cord, which then sends the message to the
brain. The brain then makes sense of that message and fires off a response. Motor
neurons distribute the instructions from the brain to the rest of your body.
A. Jangid · L. Chaudhary (B) · K. Sharma

JIET, Jodhpur, India
e-mail: laxmichaudhary1911@gmail.com
A. Jangid
e-mail: aisha.jangid@jietjodhpur.ac.in
K. Sharma
e-mail: komal.sharmaece@jietjodhpur.ac.in

https://doi.org/10.1007/978-981-15-8820-4_16
160 A. Jangid et al.
1.2 Single-Neuron Modelling
The neuron is the basic working unit of the brain, a specialized cell designed to
transmit information to other nerve cells, muscle or gland cells. Neurons are cells
within the nervous system that transmit information to other nerve cells, muscle or
gland cells. Most neurons have a cell body, an axon and dendrites. In single-neuron
modelling, we study about the behaviour and simulation of a single neuron [2].
For modelling single neurons, two types of complexity are usually associated.
Firstly, the intrinsic properties of the cell membrane that make neuronal dynamics are
so rich. Secondly, the morphology allows neurons to receive and integrate thousands
of synaptic inputs from other cells.
1.3 Development, Axonal Patterning and Guidance
Axon guidance is a subfield of neural development concerning the process by which

neurons send out axons to reach the correct targets. Axons generally follow very
precise paths in the nervous system, and finding the way so accurately is a part of
research [3].
1.4 Sensory Processing
Sensor Processing also called “sensory integration dysfunction” (SID) is the method
in which the nervous system gets the message from the sense organs and convert
those messages into suitable motor and responses for communication purpose. The
sensorimotor system includes all of the sensor, motor, and central integration and
processing components involved and also maintains joint homeostasis during bodily
movements (functional joint stability). So, we can say that functional joint stability
is basically a difficult and complex physiologic process [1].
1.5 Memory
Limbic system is the region of the brain which is deep inside the medial temporal lobe,
which includes the hippocampus, the amygdala, the cingulate gyrus, the thalamus,
the hypothalamus, the epithalamus, the mammillary body and other organs, many of
which are of particular relevance to the processing of memory. They still don’t fully
understand exactly how we remember or what occurs during recall. The process of
memory begins with encoding, then proceeds to storage and, eventually, retrieval.
Encoding is the crucial first step to creating a new memory. It allows the perceived
16 Computational Neuroscience and Its Applications: A Review 161
item of interest to be converted into a construct that can be stored within the brain, and
then recalled later from short-term or long-term memory. Encoding is a biological
event beginning with perception through the senses [3]. First, our brain consciously
registers the memory, this is called encoding. The best way to improve our memory
is to keep remembering the same thing, over and over.
1.6 Behaviours of Networks
This is called generalization. Highly complex pattern recognition can be achieved by

using a network of neurons hence the name neural networks. The networks normally
used for pattern recognition are called feed forward because they have no feedback.
They simply associate inputs with output.
1.7 Cognition, Discrimination and Learning
1.7.1 Cognition
The study drawn on the basis of aspects of psychology, linguistics, philosophy and
computer modelling gives the basis of thought, learning and mental organization. The
use of the computer as a tool for thinking how the human mind handles information
is known as the computer analogy. The computer gave cognitive psychologists an
analogy, to which they could compare human mental processing.
1.7.2 Discrimination
Discrimination, in psychology, is the ability to perceive and respond to differences

among stimuli. It is considered as a more advanced form of learning than gener-
alization, the ability to perceive similarities, although animals can be trained to
discriminate as well as to generalize (Figs. 1 and 2).
1.7.3 Learning
Discrimination, in psychology, is the ability to perceive and respond to differences

among stimuli. It is considered as a more advanced form of learning that is the ability
to perceive similarities, although animals can be trained to discriminate as well as
to generalize. Figure 3 shows Process of Cognition, discrimination and learning
[1]. Learning rule or Learning process is a method or a mathematical logic which
improves the artificial neural network’s performance by repeatedly applying over the
network.
Fig. 1 Basic neuron design

[1]
Fig. 2 Sensor processing
Fig. 3 Process of cognition,

discrimination and learning
[1]
2 Computational Neuroscience
2.1 Brain Data Recording
By the excitation of various neuron, the electric current flows in nanoamperes and
EM waves is produced and that is recorded with the use of electrodes put on the head.
The data obtained from this method is stored and processed in a special high-speed
computer. After that these data are analyzed for the various reactions of the brain for
different sensation towards body [4]. From this method, the scientist partially gets
success to record the dreams of human. They succeeded to make the blurred image
stored inside the brain which was shown to the human. After sometime they will also
achieve the success to record the dreams of a person by this method.
2.2 MRI Magnetic Resonance Imaging
The other method to analyze the brain is MRI. MRI is acronym of magnetic resonance
imaging. In this, the resonant property of hydrogen ion in brain is considered and
makes an image of the brain at higher radio frequency oscillating magnetic field [5].
2.3 Quantum Computers
A quantum computer is a device for computation that distinctively uses quantum

mechanical phenomena, such as superposition and entanglement to perform oper-
ations on data. An atom, not an electron, is the physical bit. An electron is 0 or 1.
Quantum mechanics: at atom is 0, 1 or both “coherent superposition”. The bit in
quantum mechanics is a qu bit. The brain is a complex structure so we should require
to store and process huge information at a time. The quantum computer is very fast
computers and efficient storage schemes. So, the next generation is planning to use
quantum computers for the analyzing of brain information [6]. It is a computer which
makes use of the quantum states of subatomic particles to store information. Figure 4
shows that Quantum Computer is the best way to analyze the brain.
2.4 Neural Engineering
Neural engineering (also known as neuroengineering) is a discipline within biomed-

ical engineering that uses engineering techniques to understand, repair, replace,
enhance or otherwise exploit the properties of neural systems [7].
Fig. 4 Quantum computer is best way to analyze the brain
3 Neural System as Electronic Network
3.1 Hodgkin–Huxley Model
The Hodgkin–Huxley model, or conductance-based model, is a mathematical model

that describes how action potentials in neurons are initiated and propagated. Figure 5
shows Hodgkin–Huxley Model.
Hodgkin–Huxley type models represent the biophysical characteristic of cell
membranes. The lipid bilayer is represented as a capacitance (Cm). Voltage-gated
and leak ion channels are represented by nonlinear Hodgkin–Huxley type models
represent the biophysical characteristic of cell membranes. The lipid bilayer is repre-
sented as a capacitance (Cm). Voltage-gated and leak ion channels are represented
by nonlinear (gn) and linear (gL) conductances, respectively. The electrochemical
gradients driving the flow of ions are represented by batteries (E), and ion pumps,
and exchangers are represented by current sources (Ip).
Fig. 5 Hodgkin–Huxley model

3.2 Neuron-Electronic Equivalent
We can make a neuron equivalent electronic circuits which works very similar to
neuron so that the signal transmission can travel with the speed of electron. The
neuron signal transmission speed is very slow as 12o m/s because it uses ion-exchange
method. So the electronic equivalent would be fast (Figs. 6 and 7).
The electronic neuron model is Lewis Membrane Model which is based on
Hodgkin–Huxley equation. The sodium and potassium conductance, synaptic
connections and other functions of the model are represented by discrete transis-
tors and associated components in form of parallel circuits connected between nodes
representing the inside and outside of the membrane [8].
Fig. 6 Lewis membrane model [14]
Fig. 7 Brain power processing [10]

3.3 Neural Networks
The interconnection of neurons inside the neural system is called neural networks.
The fifth generation of computers uses neural networks. A neural network is referred
as a series of algorithms that aims to understand relationships in a set of data through
a process which behaves in the way the human brain operates.
4 Biological Properties of Neural System
4.1 Genetic Algorithm
GA is a search heuristic that imitates the process of natural evolution [9]. This
heuristic is routinely used to generate useful solutions to optimization and search
problems.
4.2 DNA Related Property
DNA is the basis of life and responsible for flow of life. It stores the information of
body structure of a living being and can derive a living being correctly from a single
cell called zygote. There are four types of base in DNA such as adinine, guanine,
thymine and cytocine (A, G, T, C) [1]. A always combines with T by hydrogen bond.
This one combination can be referred to as binary ‘1’. T always combines with C by
hydrogen bond. This one combination can be referred to as binary ‘0’. There are 3.2
billions of such combinations inside the DNA of single cell of human. So, it can be
possible that some neural features of parent can undergo their offspring.
4.3 Biological Computer
Brain is a biological computer which has neurons as its functional unit. Digital bits
are only two voltages by which can store and process very less information per second
as compared to neurons. It is an analogue system. The different neuron voltages act
as several learning points and can be said bio–bits [8].
4.4 Section Division of Problems in Brain and CPU
The brain is divided into two sections as Left and Right brain which control the
right and left side of the body control, respectively. The Left side of the body control
includes the emotion, face recognition,creativity, Music/art awareness, imagination,
subjectivity, etc., whereas the right side of the body control includes the math/science
skills, written language, spoken language, objectivity, analytical, logic, reasoning,
etc.
4.5 Brain Power Processing
The brain includes the sensory processing, cognition memory action selection and
motor processing where the ‘X’ is the latent variables, ’r’ is peripheral spikes which
gives the motor actions [10]. Latent variables get encoded and are processed to give
respective motor action using peripheral spikes.
5 Applications of Computational Neuroscience
5.1 Artificial Intelligence
It is the ability of computer system that makes it able to perform tasks which normally
require human intelligence, such as visual perception, speech recognition, deci-
sion making and language translation [9, 11]. It includes knowledge representation
issues, mapping and logic by rules, probabilistic reasoning, planning system, natural
language processing, learning, artificial neural networks, Fuzzy logic & genetic
algorithms, etc.
5.2 Artificial Neural Networks
Artificial neural networks (ANNs) or connectionist systems are a computational

model used in computer science and other research disciplines, which is based on a
large collection of simple neural units (artificial neurons), loosely analogous to the
observed behaviour of a biological brain’s axons [8, 12].
5.3 Brain Like Working Machine
Artificial brain (or artificial mind) is a term commonly used in the media to describe
research that aims to develop software and hardware with cognitive abilities similar
to those of the animal or human brain. Research investigating “artificial brains” and
brain emulation plays three important roles in science. Neuroscientists are trying to
understand how the human brain works, known as cognitive neuroscience. AI has
ability to create a machine that can perofrm task similar to a human being. It is long-
term project to create machines exhibiting behaviour comparable to those of animals
with complex central nervous system such as mammals and humans, and this goal
of creating a human mimicking machine and exhibiting human-like behaviour or
intelligence is called as strong AI [11, 13].
5.4 Immortality
5.4.1 Brain Transplant (Limited Age of Brain)
A brain or whole-body transplant is a procedure in which the brain or organ of

one organism is transplanted into the body of another [6]. It is different from head
transplantation, which involves transferring the entire head to a new body unlike the
brain only. Brain data transfer from old to child and few applications such as.
Brain data storage, Living machine (One kind of computer which uses brain as
its CPU.), Living as robot-server, Cloud brain data storage.
Advanced display for blind people:
The blind people can’t see anything but due to improvements in these fields there
may be displays that can be seen by blind people, and hence they can visualize the
things through brain and take actions [5].
6 Conclusion
Computational neuroscience approaches further our understanding of brain func-

tion, and helps in translating the acquired knowledge into technological applica-
tions. From a scientific perspective discovering how the brain thinks is a major
activity in the history of mankind. Be it cellular and synaptic dynamics or biophys-
ical basis of neuronal computation or algorithms for the analysis of neuronal data,
the field has provided analytical and computational skills for understanding the
neuronal systems. The field of computational neuroscience explores the informa-
tion processing strategies employed by neural circuits in the brain. Computation”
and “information processing” are commonly used interchangeably, which presup-

poses that they are roughly synonymous terms. If that were true, computation would
entail information processing, and information processing would entail computation.
“Understanding the human mind in biological terms “has evolved as the challenging
topic for science in the twenty-first century. There is need to understand the biolog-
ical nature of perception, learning, memory, thought, consciousness and the limits
of free will, and thus it is expansion of new science of mind not from extent of how
we perceive, learn, remember, feel to a new perspective of ourselves in the context
of biological evolution. The neural science explains the behaviour in terms of the
activities of the brain.
References
1. Gazzaniga et al (2004) The cognitive neurosciences. MIT Press

2. Kriegeskorte N, Mok RM (2017) Building machines that adapt and compute like brains. Behav
Brain Sci 40
3. Grill-Spector K, Weiner KS, Kay K, Gomez J (2017) The functional neuroanatomy of human
face perception. Ann Rev Vision Sci 3:167–196
4. Mnih V et al (2015) Human-level control through deep reinforcement learning. Nature
518:529–533
5. Haynes J-D (2015) A primer on pattern-based approaches to fMRI: principles, pitfalls, and
perspectives. Neuron 87:257–270
6. LeCun Y, Bengio Y, Hinton G (2015) Deep learning. Nature 521:436–444
7. Kriegeskorte N, Kievit RA (2013) Representational geometry: integrating cognition, compu-
tation, and the brain. Trends Cogn Sci 17:401–412
8. Thomas JI (2019) Current status of consciousness research from the neuroscience perspective.
Acta Sci Neurol 2(2)
9. Dimitrov AG, Lazar AA, Victor JD (2011) Information theory in neuroscience. J Comput
Neurosci, Springer 30:1–5. https://doi.org/10.1007/s10827-011-0314-3
10. Ward J (2015) The student’s guide to cognitive neuroscience. Psychology Press
11. Frisoni GB et al (2012) N4U: expansion of neuGRID services and outreach to new user commu-
nities (poster). In: 9th e-Infrastructure concentration meeting of the european grid infrastructure,
22 Sept 2011. https://neugrid4you.eu/conferences
12. Dehaene S, Naccache L (2001) Towards a cognitive neuroscience of consciousness: basic
evidence and a workspace framework. Elsevier
13. Redolfi A, McClatchey R et al (2009) Grid infrastructures for computational neuroscience: the
neuGRID example. Future Neurol 4(6): 703–722. https://doi.org/10.2217/fnl.09.53
14. Sejnowski TJ, Jolla L, Chase C (2015) Computational neuroscience. In: International
encyclopedia of the social &behavioral sciences, 2nd edn, pp 480–484, Elsevier
Chapter 17
Optimization of Band Notch
Characteristic in Ultra-Wideband
Microstrip Patch Antenna for Wireless
Power Transfer
Ashish Mathur, Geetika Mathur, and Abhijit kulshrestha
1 Introduction
Federal communications commission (FCC) allocated a block of radio spectrum

from 3.1 to 10.6 GHz for UWB operations [1]. UWB systems can support more
than 500 Mbps data transmission within 10 m [1]. Compact size, low-cost printed
antennas with Wideband and Ultra-wideband characteristic are desired in modern
communications. The Ultra-wideband antennas can be classified as directional and
omni-directional antennas [2]. A directional antenna has high gain and is relatively
large in size. It has narrow field of view. Whereas the omni-directional antenna has
low gain and relatively small in size. It has wide field of view as it radiates in all the
directions [2]. The UWB antennas have broad band. There are many challenges in
UWB antenna design. One of the challenges is to achieve wide impedance bandwidth.
UWB antennas are typically required to attain a bandwidth, which reaches greater
than 100% of the centre frequency to ensure a sufficient impedance match is attained
throughout the band such that a power loss less than 10% due to reflections occurs at
the antenna terminals. The bandwidth of the microstrip antenna can be enhanced by
modifying the ground plane [3]. Many designers have tried various ways to improve
the structure of the traditional circular antennas, and many valuable results have been
obtained [4–8].
A. Mathur (B) · A. kulshrestha

JIET Group of Institution-Jodhpur, Jodhpur, Rajasthan, India
e-mail: ashish.mathur@jietjodhpur.ac.in
A. kulshrestha
e-mail: abhijit.kulshrestha@jietjodhpur.ac.in
G. Mathur
Jagannath University Jaipur, Jaipur, Rajasthan, India
e-mail: geetika.mathur@jietjodhpur.ac.in

https://doi.org/10.1007/978-981-15-8820-4_17
172 A. Mathur et al.
2 Antenna Configuration and Design
For the calculation of length and width of the patch antenna, we used basic formulas
for length and width of patch. The ground plane is modified to enhance the bandwidth
of the antenna. The proposed antenna designed on a FR4 substrate with dielectric
constant εr = 4.4 and height of the substrate is h = 1.6 mm. The substrate has length
L = 50 mm and width W = 50 mm. The substrate is mounted on ground of 18 mm
length and 50 mm width (Figs. 1 and 2).
The proposed design is capable of passing four bands in the range of 2.1–3.25 GHz
in the range of ISM (2.4–2.4835 GHz), Bluetooth (2.4–2.484 GHz) and rejection of
Wi max IEEE 802.16 (3.3–3.7 GHz) band at absolute bandwidth in GHz below return
loss of −10 dB is 2.1 to 3.25 GHz = 1.25 GHz, Second 5.2 to 6.2 GHz = 1.0 GHz,
third 7.8 Ghz to 9.0 GHz and fourth 9.10 to 11.1 GHz. This antenna is resonant at
four centre frequencies. First is 2.5 GHz with absolute bandwidth 1.0 GHz, second is
5.5 GHz with absolute bandwidth 2.0 GHz, third is 8.5 GHz and fourth is 10.5 GHz
for UWB applications.
Fig. 1 UWB spectral mask per FCC (Modified) part 15 rules [1]
Fig. 2 Geometry of rectangular patch top view & bottom view

17 Optimization of Band Notch Characteristic in Ultra-Wideband … 173
3 Simulation Results
Figures 3 and 4 show parametric study of S11 parameter for multiband patch antenna
with optimized ground length Lg = 18 mm. This antenna is suitable for oper-
ating frequency of 2.1–3.25 GHz, 5.2–6.2 GHz, 7.8–9.0 GHz and 9.10–11.1 GHz
in UWB. It is shown that return loss of the antennas is better than −10 dB. The
VSWR obtained is less than 1.5 and the patch antenna is found to have the compact
size. The return loss value of first, second, third and fourth band is −16 dB,−45 dB,
−24 dB and −20.1 dB, respectively.
Figure 5 shows the relationship between VSWR with frequency for the proposed
design. In this, the value of VSWR is ≤2 for four different centre frequencies. First
is 2.5 GHz with absolute bandwidth 1.0 GHz and second is 5.5 GHz with absolute
bandwidth 1.0 GHz and third is 8.5 GHz with absolute bandwidth 1.2 GHz and fourth
is 10.5 GHz with absolute bandwidth 2.1 GHz for UWB applications.
Fig. 3 S11 of Patch antenna with different ground plane effects Lg
Fig. 4 S11 of patch antenna with Lg = 18 mm

Fig. 5 VSWR of multiband notch patch antenna
The Plot curve of Directivity, Gain in 3D Polar are shown in Figs. 6 and 7, Radia-
tion efficiency and radiation pattern are shown in Figs. 8 and 9, the current distribution
of the proposed design is shown in Fig. 10. The simulated values of directivity are
3.65 dB with 2.64 dB antenna gain and 82% radiation efficiency is calculated for the
proposed geometry. The uniformly current distribution and bidirectional radiation
pattern are obtained for proposed geometry at 7 GHz.
Ground—Plane/Substrate—Related effect: The three important points can be
observed after the parametric analysis of ground plane with FR4 substract. First,
it is seen that the impedance matching is very sensitive to the feed gap, especially
Fig. 6 Directivity of patch antenna at 7 GHz

Fig. 7 3D Polar gain of patch antenna at 7 GHz
Fig. 8 Radiation efficiency at 7 GHz

Fig. 9 Radiation patten at 7 GHz
Fig. 10 Measurement of E field distribution at 7 GHz

Fig. 11 Measurement of H field distribution at 7 GHz
at higher frequencies. Second, the length of the ground plane affects the impedance
matching more significantly at higher frequencies than at lower frequencies [9]. This
finding is consistent with the current distributions where more current is concentrated
on the ground plane at the higher frequencies than at lower frequencies [10]. Last,
the impedance response is also affected by the dielectric constant [11]. In this study,
a change in the dielectric constant leads to a shift in the characteristic impedance of
the feeding strip from 50 (Fig. 11).
4 Fabrication, Measurement Setup and Testing
The antenna structure is fabricated on FR 4 substrate using Photolithography tech-

nique. The proposed design is tested on vector network analyzer. The top view and
measurement setup of fabricated antenna are shown in Figs. 12 and 13 (Table 1).
The measured result of S11 for the proposed design is calculated by vector network
analyzer and on the basis of measured results, we conclude that this antenna is suitable
for frequency band of 7.4–10.2 GHz with resonant frequency at 8.25 GHz with good
return loss and VSWR values (Figs. 14 and 15).
Fig. 12 Fabricated design of proposed antenna
Fig. 13 Measurement setup of S11 at 7 GHz
Table 1 Antenna designing parameters

W sub L sub Wg Lg Ws Ls Wp Lp
50 50 50 18 12 4 30 30
Fig. 14 Measurement setup of VSWR at 7 GHz
Fig. 15 Measurement setup of Input Impedance at 7 GHz
5 Conclusion
In this paper, multiband patch antenna with band Notch Characteristic in UWB is
simulated using HFSS-13. The proposed antenna has advantages of small size, easy
fabrication and simple construction. The simulated results of the proposed antenna
show that return loss is less than −10 dB and VSWR is less than 1.5. The measured
results of this antenna show that the antennas can be good candidates for the four
operating frequency of 2.3–3.2 GHz, 5.4–6.2 GHz, 7.4–9.0 GHz and 9.1–10.2 GHz
with four resonant frequencies and good return loss values. The gain of antenna is
2.64 dB and radiation efficiency 82% calculated. Microstrip line feeding is used for
transmission of EM wave.
Acknowledgements The authors thank Prof. (Dr.) J. P. Agarwal Department of Electronics &
Communication Engineering GIT Jaipur, India and Prof. R.K. Malviya Sectary ATMS for providing
antenna design, fabrication and characterization facilities.
References
1. Ashtankar PS, Dethe CG (2012) Design and modification of circular monopole UWB antenna
for WPAN application. Comput Eng Intell Syst ISSN 2222-1719 (Paper) ISSN 2222-2863
(Online) 3(5)
2. Barba Qinrino S (2010) UWB circular slot antenna provided with an inverted-l notch filter for
the 5Ghz WLAN band. Progr Electromag Res PIER 104, 1–13
3. Chang CC, Watanable F, Inamura H (2006) Potential of UWB technology for the next gener-
ation wireless communications. In: IEEE ninth international symposium on spread spectrum
techniques and applications, pp 422–429
4. Behdad N, Sarabandi K (2005) A compact antenna for Ultrawide- Band applications. IEEE
Trans Antennas Propag 53(7)
5. Balanis C (1997) Antenna theory: analysis and design. John Wiley & Sons Inc, New York
6. Hassanien MA, Hamad EKI (2010) Compact rectangular u-shaped slot microstrip patch antenna
for UWB applications. In: IEEE APS, middle east conference on antennas and propagation
(MECAP), Cairo, Egypt
7. Bhomia Y, Kajla A, Yadav D (2010) V-slotted triangular microstrip patch antenna. Int J Electron
Eng 2(1): 21–23
8. Kaur N, Sharma N, Singh N (2017) A study of different feeding mechanisms in microstrip
patch antenna. Int J Microwave Appl 6(1)
9. Chen ZN, See SP, Qing X (2007) Small printed ultrawidwband antenna with reduced ground
plane effect. IEEE Trans Antennas Propag 55(2)
10. Xi D, Wen LH, Yin YZ, Zang Z, Mo YN (2010) A compact dual inverted c-shaped slots antenna
for WLAN applications. Progr Electromag Res Lett 17:115–123
11. Prombutr N, Kirawanich P, Akkaraekthal P (2009) Bandwidth enhancement of UWB micro
strip antenna with a modified ground plane. Int J Microwave Sci Technol
Chapter 18
Experimental Investigation
for Energy-Conscious Welding Based
on Artificial Neural Network
Sudeep Kumar Singh, Suvam Sourav Swain, Amit Kumar,

Prashanjeet Patra, Nitesh Kumar, and A. M. Mohanty
1 Introduction
Shielded Metal Arc Welding (SMAW) of Mild Steel (MS) finds wide application
in structural frames, pipelines, visually aesthetic designs, and repair due to its high
ductility and weldability properties [1–3]. Welding remains the most widely adopted
joining process in the industry despite its high energy-intensive property. The selec-
tion of proper welding parameters is very important in a multi-input multi-output
process like welding [4, 5]. The mechanical properties of welded joints largely
depend on process parameters used in the manufacturing process [6]. The welder
generally focuses on the quality aspects of the produced joints and pays lesser atten-
tion to the process parameters. In practice, the welding process parameters selected
imparts a large influence on the resources consumed like joint quality, percentage of
rework/rejection, and energy consumed [7]. Power consumption is one among many
factors responsible for the negative environmental effects generated from welding
operation, raising the need for characterization of the SMAW process considering
sustainability aspects [3, 8]. Thus, the present study intends to draw a relationship
between the four influential input parameters and the four output parameters adopted
for investigation.
S. K. Singh (B) · S. S. Swain · A. Kumar · P. Patra · N. Kumar · A. M. Mohanty

Centurion University of Technology and Management, Odisha Bhubaneswar, India
e-mail: sudeepuce008@gmail.com

https://doi.org/10.1007/978-981-15-8820-4_18
182 S. K. Singh et al.
2 Literature Survey
Adnan et al. [4] carried out Pareto Analysis to find uncontrollable input parameters of
the GMAW welding process. They developed three different ANN models for input,
output parameter prediction, and classifying products. ANN was also employed for
investigating the effects of process parameters in laser welding of AA5754 aluminum
alloy [9]. Two parameters welding speed and shielding gas were varied, and the opti-
mization process was implemented using an Excel add-in named Neural Tools. In
yet another study, authors developed two different ANN models one for classifica-
tion of defective products and other for prediction of input parameters [5]. Welding
processes have a poor environmental image for which optimization of key welding
parameters is very crucial. A hybrid approach involving neural network and fuzzy
logic is used for optimizing SMAW process parameters from the sustainability point
of view [8]. Current, voltage and welding speed are considered for analysis. Welding
of dissimilar metals involving Al alloy and stainless steel has been studied using the
laser-arc welding technique [10]. Taguchi is used for studying the effect of various
welding parameters to get optimum parameters of angular distortion in SMAW [11].
TIG welding parameter has been optimized using Response Surface Methodology
(RSM), central composite design on mild steel [12], and grey wolf optimizer [13] on
high strength low alloy 15CDV6 steel. RSM has also been adopted for optimizing
GMAW parameters for welding mild steel IS:2062 [14]. Authors [15] have devel-
oped model for prediction of mechanical and microstructural properties of copper
plate welding using Friction Stir Welding. RPLNN and GA have been used involving
three inputs and two response parameters.
Weld quality considering tensile properties and microstructure were analyzed
based on power distribution using an arc assisted fiber laser welding of Al–Mg alloy
[16]. Tensile and impact properties in multi-pass SMAW have been investigated by
Saxena et al. for determining the influence of welding consumables in Armox 500T
alloy [17]. Mechanical properties and microstructure of MS welded parts under
varying current were analyzed using the E7016 electrode [1]. The highest tensile
strength was obtained at 75A with minor welding defects. Sheets of different thick-
nesses welded using SMAW and GMAW were investigated for finding a new set of
welding parameters for structural grade steel welding [6].
The main aim of the current research work is to study the influence of varying
input parameters on the output quality of the joint.
The arrangement of the paper is as follows. The experimental methodology is
explained in Sect. 3. The next section discusses the outcomes of the experimental and
test results. The fifth section discusses the application of ANN for welding parameter
selection. The sixth section presents conclusions obtained from the analysis and also
provides directions for future scope.
18 Experimental Investigation for Energy-Conscious Welding … 183
3 Methodology
The strategy followed in the current work can be divided into different sections of
which, arc welding, testing for obtaining output data, and selection of input parame-
ters to the welding process based on influential responses of welding are important.
Arc welding of mild steel considering energy consumption has been considered in the
present investigation. The strategy followed in the current investigation is presented
pictorially in Fig. 1.
Mild steel plates of different thicknesses 3, 5, and 10 mm (three levels) were
utilized in the welding process. The welding parameters, current, joint gap, and
face width were also varied during the experiment. The input parameters considered
in the investigation include the welding parameters and the plate thicknesses. The
output parameters considered are Ultimate Tensile Strength (UTS), impact energy
(Izod), Rockwell hardness, and energy consumption. The input parameters (factors)
involved in the study are presented in Table 1.
Mild steel procured in flat form was first cut to a rectangular shape with length
200 mm and width 100 mm. One longitudinal edge of each plate was beveled to
produce a double V-groove butt joint. The including angle of the V-shaped joint is
60° for all the plates used. The chemical composition of the plates was tested using
XRF spectrometer, and the obtained values are tabulated in Table 2.
The data presented in the table displays close conformance in terms of composition
for both the workpiece and filler metals. The filler rod used in the welding process is
3.15 mm in diameter Superweld E6013 manufactured by ESAB. The XRF samples
for both material types were prepared by grinding on a surface grinder.
Fig. 1 Experimental methodology
Table 1 Different values for the input variables

Sl. no Factors Level 1 Level 2 Level 3
1 Current 100A 110A 120A
2 Plate thickness 3 mm 5 mm 10 mm
3 Root gap 0 mm 1 mm 2 mm
4 Face width 0 mm 1 mm 2 mm
Table 2 Material composition

Sl. no. Base material Si Mn S P Fe
1 Mild steel plate 0.720 0.709 0.132 0.029 96.840
2 Electrode E6013 1.451 0.437 0.125 0.034 96.115
Table 3 Experimental values in the investigation

Sl. no. Current Plate Root gap Face Power UTS Hardness Impact
(A) thickness (mm) width (kW) (MPa) (HRB) energy
(mm) (mm) (J)
1 100 3 0 0 4.73 481 76.4 60
2 100 5 1 1 4.52 411 77.25 62
3 100 10 2 2 5.32 305 83.9 74
4 110 3 1 2 4.59 295 78.1 50
5 110 5 2 0 5.52 501 78.6 52
6 110 10 0 1 5.14 406 85.55 160
7 120 3 2 1 5.88 458 80.6 52
8 120 5 0 2 6.59 362 84.05 112
9 120 10 1 0 5.89 329 82.65 110
The plates were cleaned properly using solvent to remove all dirt, rust present on
the surface of the material to be welded. It is followed by welding the plates using
process parameters obtained from TAGUCHI orthogonal array design presented in
Table 3.
3.1 The Welding Process
Similar to the raw material of three different thickness values, the input current has
also been varied into the same number of current values and adopted for the exper-
iments: 100, 110, and 120 amperes. The remaining two input variables adopted are
root gap and face width. Three different values were considered for both the vari-
ables as 0, 1, and 2 mm. All the varying parameters are taken together, including the
plate thickness values, make the total number of factors involved in the experimental
design as four. The number of levels for each factor is three. Thus, if the full factorial
design of experiments was to be considered, the total number of experiments would
become 27. To reduce the number of experiments, Taguchi Design of Experiment
(DoE) method was adopted. Using L9 Taguchi orthogonal array design adopting a
four-factor and three-level experimental approach, the total number of experimental
runs was reduced to 9. The experimental design adopted for the experiments is
presented in Table 3.
The welding process was carried out by using RS400 a Thyristorised MMA
welding machine manufactured by ESAB India Ltd. The machine is equipped with
50 Hz 3-phase power supply with an input voltage of 415 volts and 27-ampere current.
The welding runs were carried out using the AC power supply.
A 3-phase power analyser, model no DPATT-3Bi, manufactured by Uma Elec-
tronics Enterprises, Jaipur, India, was used for measuring the instantaneous power
consumption values during the arc welding process. A three-phase four-wire
connection was used in the process of measurement.
Table 3 presents the four factors and the values of the three levels of process param-
eters adopted in the experimental runs. It displays the values of different process
parameters used in the welding process. Four different parameters welding current,
plate thickness, root gap, and face width are used for designing nine number of
experiments in total.
The welding speed was considered constant throughout the experiment. The plates
of 3 mm thickness were welded using a single pass of welding, but multiple runs were
necessary for plates with 5 mm and 10 mm thickness. The former was welded with
two passes, and for the latter three number of welding, passes were used. In total,
nine numbers of welding joints were produced and processed further for preparing
test samples for tensile. Rockwell and Izod impact tests to be conducted further. The
details of the test procedure and results have been explained in the next section.
4 Post-weld Testing
The welded steel plates were cleaned to remove the slag deposited during welding by
using a chipping hammer and wire brush. Tensile, hardness, and Izod test specimens
were extracted from the welded plates of different thicknesses with the respective
dimensions, presented in Fig. 2.
Welding beads were removed by grinding operation from the welded surface for
both the tests.
The tensile test was conducted on a Universal Testing machine manufactured
by Blue Star Engineering & Electronics Ltd., having a maximum capacity of 1000
kN. The test specimens were made to undergo the tensile testing procedure, and the
Ultimate Tensile Strength values for each test specimen were noted. The average
value of HRB was calculated after measuring hardness values at two different points
on the weld bead surface. The samples prepared for the Izod test were carried out
using Impact test machine, and values of energy absorbed before failure for individual
specimen were recorded. The values of UTS, HRB, and energy absorbed have been
presented in Table 3 under respective columns.
Figure 3 displays phases of sample preparation for different tests after conducting
the tensile, Rockwell, and Izod tests. The Fig. 4a displays the Impact testing machine,
and 4 (b) depicts the Rockwell hardness testing machine used for the experimentation.
Fig. 2 Schematic diagrams of a Tensile Test specimen; b Izod specimen
5 Parameter Selection Using ANN
Neural networks find a wide application and recognized as efficient solvers of non-
linear problems. Successful applications have been reported in literature containing
real-world problems. Thus, ANN has been selected for finding optimum input param-
eters for SMAW in the present study. The architecture for the employed neural net
is presented in Fig. 5.
Fig. 3 Pictorial representations of a Tensile Test; b Hardness; c Izod Specimens
Fig. 4 a Impact testing machine; b Rockwell hardness testing machine
An Artificial Neural Network was modeled for training using the data collected
from the conducted experiments. The Bayesian Regularization backpropagation
method is used for the construction of the network. This method is generally used
for difficult, small, and noisy datasets (Fig. 6).
In the current construction, the data set is small and prone to noise in the measured
value; thus, the application of Bayesian Regularization fits our requirement. ‘trainbr’
learning function is used in the Matlab R2019a platform. The network takes 70% of
Fig. 5 ANN architecture
data for training, 15% for validation, and 15% for testing. The ANN model developed
in this study involves an input layer, one hidden layer, and one output layer. The
input layer consists of four neurons; each neuron corresponding to individual input
parameters and the output layer containing four neurons, representing one output
parameter each. The hidden layer employs 50 neurons. The most promising network
architecture is based on the trial and error method for which many trials have been
conducted to arrive at the best combination. The performance of the network has
been discussed in detail in the conclusion section.
6 Conclusion
The current work involves four input and four output variables for SMAW welding of
structural grade mild steel. The quality of the welding has been tested by measuring
UTS of the welded joint by applying load in the transverse direction, measuring the
impact energy absorbed by the joint before failure, hardness on the bead surface, and
also the power consumed for joint preparation. The input and output were fed into an
ANN network suitably designed for the purpose. The modeled network is capable of
selecting all the four types of input parameters considered in the present work based
on desired values of output parameters like energy consumed, UTS hardness, and
impact energy. This work can be extended to other welding methods. Other crucial
variables not considered in the present work may be considered as future research
scope.
Fig. 6 a Regression; b error histogram; c performance; d training progress; e training state
References
1. Faqih IA, Ma’arif S, Sukarjo H (2019) The effect of current variation on mma welding to
mechanical properties and microstructure of mild steel
2. Ahmed AN, Noor CM, Allawi MF, El-Shafie A (2018) RBF-NN-based model for prediction of
weld bead geometry in Shielded Metal Arc Welding (SMAW). Neural Comput Appl 29(3):889–
899
3. Alkahla I, Pervaiz S (2017) Sustainability assessment of shielded metal arc welding (SMAW)
process. In: IOP conference series: materials science and engineering, IOP Publishing,
244(1):012001
4. Aktepe A, Ersöz S, Lüy M (2014) Welding process optimization with artificial neural network
applications. Neural Netw World 24(6):655–670
5. Aktepe A, Ersöz S, Lüy M (2012) Backpropagation neural network applications for a welding
process control problem. In: International conference on engineering applications of neural
networks. Springer, Berlin, Heidelberg, pp 172–182
6. Khamari BK, Dash SS, Karak SK, Biswal BB (2019) Effect of welding parameters on mechan-
ical and microstructural properties of GMAW and SMAW mild steel joints. Ironmaking
Steelmaking:1–8
7. Singh SK, Samal BK, Pradhan SR, Ojha SR, Saffin MD, Mohanty AM (2019) Sustainable
analysis of TIG parameters for welding aluminum alloy considering joint gap and welding
current. In: International conference on application of robotics in industry using advanced
mechanisms. Springer, Cham, pp 316–323
8. Vimal KEK, Vinodh S, Raja A (2017) Optimization of process parameters of SMAW process
using NN-FGRA from the sustainability view point. J Intell Manuf 28(6):1459–1480
9. Casalino G, Facchini F, Mortello M, Mummolo G (2016) ANN modelling to optimize
manufacturing processes: the case of laser welding. IFAC-PapersOnLine 49(12):378–383
10. Gao M, Chen C, Mei S, Wang L, Zeng X (2014) Parameter optimization and mechanism of
laser–arc hybrid welding of dissimilar Al alloy and stainless steel. Int J Adv Manuf Technol
74(1–4):199–208
11. Arifin A, Gunawan AM, Yani I, Pratiwi DK, Yanis M, Sani KA (2019) Optimization of Angular
Distortion on Weld Joints Using Taguchi Approach. Jurnal Kejuruteraan 31(1):19–23
12. Srivastava S, Kumar S, Garg RK (in press) A multi-objective optimisation of TIG welding
parameters using response surface methodology
13. Skariya PD, Satheesh M, Dhas JER (2018) Optimizing parameters of TIG welding process
using grey wolf optimization concerning 15CDV6 steel. Evol Intel 11(1–2):89–100
14. Srivastava S, Garg RK (2017) Process parameter optimization of gas metal arc welding on IS:
2062 mild steel using response surface methodology. J Manufact Process 25:296–305
15. Azizi A, Barenji A, Barenji R, Hashemipour M (2016) Modeling mechanical properties of FSW
thick pure copper plates and optimizing it utilizing artificial intelligence techniques. Sensor
Netw Data Commun 5(142):2
16. Leo P, Renna G, Casalino G, Olabi AG (2015) Effect of power distribution on the weld quality
during hybrid laser welding of an Al–Mg alloy. Opt Laser Technol 73:118–126
17. Saxena A, Kumaraswamy A, Reddy GM, Madhu V (2018) Influence of welding consumables
on tensile and impact properties of multi-pass SMAW Armox 500T steel joints vis-a-vis base
metal. Defence Technol 14(3):188–195
Chapter 19
Detection of GSM Signal Using Energy
Detection and Matched Filter-Based
Techniques
Bablu Kumar Singh and Sanjay Bhandari
1 Introduction
The increase in demand of mobile communication and its reach to almost every
person the bandwidth requirement has also increased, and current spectrum alloca-
tion technique is being over utilized in some areas and underutilized in other areas.
To detect the presence of signal in GSM band, we have used energy detection and
matched filter detection method, and using Cognitive Radio (CR) techniques the
spectrum which are unused can be used to increase the spectrum efficiency. Cogni-
tive Radio enables the user to determine the presence of primary user/signal, over
available spectrum, if Primary User (PU) is not present then the available spectrum is
considered as white spaces which can be made available for unlicensed users. To allo-
cate the white spectrum for secondary users using spectrum division and spectrum
mobility for spectrum utilization [1–3]. Spectrum Sensing is the most important and
critical task to establish cognitive radio networks for spectrum utilization in GSM
network [4–6].
B. K. Singh
MBM Engineering College, JNV University, Jodhpur, Rajasthan, India
e-mail: bablukumar.singh@gmail.com
S. Bhandari (B)
Jodhpur Institute of Engineering and Technology, Jodhpur, Rajasthan, India
e-mail: sanjaybhandari41@gmail.com

https://doi.org/10.1007/978-981-15-8820-4_19
192 B. K. Singh and S. Bhandari
2 Spectrum Sensing
To enhance the detection probability in the available spectrum, Spectrum detection

techniques are used. Sensing is done across Geographical Space, Frequency, Time,
Phase, and code. The analysis of spectrum is based on several factors such as radio
frequency spectrum used by mobile devices, network conditions, and communica-
tion protocol. The objective of Spectrum Sensing [6, 7] is to detect the presence of
transmissions from primary users over an assigned channel in GSM network. If PU
is absent, then how secondary users use the primary spectrum without disrupting the
communication of primary user. This can only be possible if by some method we
can detect the channel and identify signal is present on channel or not. To detect the
presence of signal, we have considered two methods [8] in this paper. After detec-
tion and analysis of presence of signal characteristics, the channel characteristics of
GSM can be determined which can be utilized for channel optimization using Store
Forward Base Trans-receive System [9].
2.1 Energy Detection Method
Energy detection is a method which is used to detect the presence of signal over
assigned spectrum. It is easier method because it does not require prior knowledge
of type of signal available on channel. It is based on the power measurement of the
received signal. The signal block for energy detection is shown in Fig. 1
In Fig. 1, the threshold decision may be made on two hypotheses for detection of
primary and secondary user signals, first hypothesis is H0 and the other hypothesis
is H1 . H0 is the case when signal is absent, and H 1 describes the case where signal is
present. And by measuring the energy we can estimate hypothesis to decide whether
H0 or H1 is correct. In energy detection, we have two general performance matrices
which are used to evaluate the performance [8, 10, 11]. The hypothesis considered
for analysis of presence/absence of primary users are H 0 and H 1 .
H 0 : Primary user is absent
y(n) = x(n)n = 1, 2, 3 . . . N (1)
H 1 : Primary user present
Fig. 1 Energy detector block

19 Detection of GSM Signal Using Energy Detection … 193
y(n) = x(n) + u(n)n = 1, 2, 3 . . . N (2)
where u (n) is noise and x(n) is the primary signal.

Sometimes the signal is detected and sometimes the noise acts like signal but it
is very important to identify the signal which may be considered as probability of
detection and probability of false alarm. The probability of detection Pd is defined
by
Pd = Pr (Y > λ/H1 ) (3)
And the probability of false alarm P f is defined by
P f = Pr (Y > λ/H0 ) (4)
where λ is threshold value.
2.2 Matched Filter Detection Method
This method is a detection technique which performs a coherent detection of the

primary signal [6]. Spectrum sensing using matched filter requires synchronization
with the primary system and also able to demodulate the primary signal, i.e., the
prior information about the primary system must be known to secondary sensing
user like preamble signaling, channel estimation, modulation type of signal being
transmitted over the GSM network. To detect signals with maximum Signal to Noise
Ratio (SNR), matched filter detection techniques may be used at receiver. Matched
filter is a linear filter that works on the phenomena to maximize the output signal
to noise ratio and can be applied to cognitive radio user having evidence about the
primary signal. Then, this detection method may be expressed as [8]
∞

Y [n] = h[n − k]x[k] (5)
K =−∞
where x is unknown signal convolved with h, the impulse response of matched filter.
Detection by using matched filter is useful only when the information about the
source signal is known.
3 Simulation and Results
Simulation provides interactive access to check the performance and comparative

analysis for energy detection and matched filter detection method theoretically as
well as practically by varying various parameters. The sample data is collected from
USRP2920 in GSM band, and this data is being used for analysis. The detection
techniques are applied on the received signal and processed data using MATLAB
and are compared on the basis of probability of detection and probability of false
alarm with AWGN channel. The results are shown with different SNR value. The
change in observation of probability of detection and probability of false alarm is
shown in Fig. 1.
The analytical and simulated performances of energy detection are shown in
Fig. 2 with SNR with detection probability and varying probability of false alarm.
The simulated result displays that with the increase in probability of false alarm from
10–5 to 10–1 , the probability of detection raises, approaching a similar tendency to
the simulation. At high SNRs, the effect of probability of false alarm is negligible as
the detector can achieve high detection probability. This is observed near 5 dB SNR,
where the probability of finding the probability of false alarm values under consider-
ation is at least 0.9. As the SNR declines, the effect of the various probability of false
alarm values becomes easily distinguishable, i.e., at –5 dB SNR, the probability of
detection is 0.4 for the probability of false alarm of 10–1 and 0.12, for the probability
of false alarm of 10–2 .
Detection probability with varying SNR value can be observed in Fig. 3, which
shows that as the SNR increases, the detection probability increases and the proba-
bility of false alarm decreases it means we should keep high SNR for detection of
signal.
Fig. 2 Performance of energy detection (SNR, Vs, PD under varying Pfa )

Fig. 3 Performance of energy detection (PFA , Vs, PD under varying SNR)
The analytical performance match filter in Fig. 4 shows that as we increase the
probability of false alarm increases from 10–5 to 10–1 , the probability of detection
increases, approaching an alike tendency to the simulation performance.
Fig. 4 Performance of matched filter detection (SNR, Vs, PD under varying Pfa )
Fig. 5 Performance of matched filter detection (PFA, Vs, PD under varying SNR)
The performance of match filter is shown in Fig. 5, at high SNR the effect of
probability of false alarm is maximum. This can be observed at 5 dB SNR. The
probability of detection for the probability of false alarm values under consideration
is varying from 0.01 to 0.7. As the SNR declines, the effect of the various probabilities
of false alarm values is lowest.
The comparative graph between energy detection and match filter detection is
shown in Fig. 6, and it is evident from the graph that energy detection performance
is poor at low SNR, it requires a minimum SNR for its working. The result shows
Fig. 6 Energy detection versus matched filter under varying PFA

that energy detection starts working at –10 dB of SNR in our simulated result while
matched filter detection performance is better under lower SNR values.
4 Conclusion
The performance of energy detection and matched filter increases with increasing
SNR as the probability of detection of signal increases and correspondingly the
probability of false detection decreases. At lower SNR, matched filter performance
is better than energy detection (Below –5 dB). It is evidenced from Fig. 6 that with
increase in SNR value the performance of energy detection improves over matched
filter and for lower value of SNR matched filter method performance is improved
compared to energy detection method.
References
1. Ali et al (2017) Advances on spectrum sensing for cognitive radio networks: theory and
applications. IEEE Commun Surv Tutor 19(2):1277–1304
2. Dobariaa A, Sodhatarb S (2015) A literature survey on efficient spectrum utilization: cognitive
radio technology. Int J Innov Emerg Res Eng 2(1)
3. Shahzad A et al (2010) Comparative analysis of primary transmitter detection based spectrum
sensing techniques in cognitive radio systems”, Aust J Basic Appl Sci 4522–4531. INS Inet
Publication
4. Haykin S, Thomson DJ, Reed JH (2009) Spectrum sensing for cognitive radio. IEEE Proc
97(5):849–877
5. Arjoune Y, Mrabet ZE, Ghazi HE, Tamtaoui A (2018) Spectrum sensing: enhanced energy
detection technique based on noise measurement. In: IEEE 8th annu comput commun workshop
conf (CCWC), Las Vegas.
6. Abbas N, Nasser Y, Ahmad KE (2015) Recent advances on artificial intelligence and learning
techniques in cognitive radio networks. EURASIP J Wirel Commun Network.
7. Masonta MT, Mzyece M, Ntlatlapa N (2013) Spectrum decision in cognitive radio networks:
a survey. IEEE Commun Surv Tutor 15(3):1088–1107, 3rd Quart.
8. Salama U, Sarker PL, Chakrabarty A (2018) Enhanced energy detection using matched filter
for spectrum sensing in cognitive radio networks. In: Joint 7th international conference on
informatics, electronics & vision (ICIEV) and 2018 2nd international conference on imaging,
vision & pattern recognition (icIVPR), Kitakyushu, Japan, pp 185–190
9. Singh BK, Bhadada R (2020) Enhancing spectral efficiency of mobile communication by store
forward base transceiver system, 2020 7th Int Conf Signal Process Integrated Networks (SPIN),
Noida, India, 2020, pp. 781–784, http://doi.org/10.1109/SPIN48934.2020.9071058
10. Talukdar J, Mehta B, Aggrawal K, Kamani M (2017) Implementation of SNR estimation based
energy detection on USRP and GNU radio for cognitive radio networks. In: IEEE WiSPNET
11. Eduardo AF, Caballero RGG (2015) Experimental evaluation of performance for spectrum
sensing: matched filter vs. energy detector. In: IEEE Colombian conference on communication
and computing, Popayan, Colombia
Chapter 20
Simulation of Performance
Characteristics of Different PV Materials
Harish Kumar Khyani and Jayashri Vajpai
1 Introduction
The research in the area of generation of solar electricity is motivated by both

improvement of the performance for different Photovoltaic (PV) materials and by
a decrease in the technological realization cost. Study and development of new
photovoltaic materials is an emerging area of research. The variety of photovoltaic
semiconductor materials, developed over the past two decades, possess different
optical, electrical, and mechanical properties. However, the deployment of these
new materials and their associated design technologies for manufacturing solar cell
depend essentially on the availability of material and often impose an elevated price.
These different materials are characterized by their typical properties, which can
considerably influence the price and the performance of the variety of solar cells
manufactured by employing them. The present stake in the growth of photovoltaic
systems is to find a compromise between the cost and performance to guide the
selection of photovoltaic material for designing modules for bulk energy genera-
tion. MATLAB/Simulink-based modeling of photovoltaic module has already been
accomplished by the authors [1] in a previous work. This model can be adapted to
represent different types of photovoltaic materials by changing certain parameters.
The simulation of performance features of six commonly used different types of
photovoltaic materials, and their comparative study is offered in this paper. This
paper is divided into five sections. The next section discusses the PV module tech-
nologies being used presently. The third section presents MATLAB/Simulink-based
model of PV system followed by the simulation of different PV materials. Finally,
H. K. Khyani (B) · J. Vajpai

M.B.M Engineering College, JNVU, Jodhpur, India
e-mail: khyani.harish@gmail.com
J. Vajpai
e-mail: jvajpai@gmail.com

https://doi.org/10.1007/978-981-15-8820-4_20
200 H. K. Khyani and J. Vajpai
the comparison of the performance characteristics of these models has been carried
out before concluding comments.
The research for mathematical modeling of PV cells was initiated by physicists
in the year 1883 but engineering-oriented modeling began quite recently around
the year 2000. With the emergence of computational modeling techniques along
with increased interest of researchers toward simulation-based studies, the use of
MATLAB software has become the ruling trend in this research area. Tsai et al.
[2] pioneered the progress of generalized MATLAB/Simulink® model for solar PV
cell to examine the consequence of solar irradiance and cell temperature. This was
followed by a circuit-based simulation model designed by González-Longatt [3], in
order to evaluate the electrical behavior of the cell with respect to changes in tempera-
ture and irradiance. Further, Bhatt and Thakker demonstrated MATLAB-based simu-
lation of photovoltaic array at different temperatures to obtain their electrical charac-
teristics as a function of temperature [4]. Yousef Mahmoud, Xiao, and Zeineldin have
simplified the model and extended it for the PV modules while employing precise
estimates of the model parameters directly from manufacturer datasheets [5]. Panwar
and Saini have studied the problem of model parameter determination based on the
four-parameter model using MATLAB/Simulink [6].
These researchers simulated MATLAB-based models of photovoltaic modules
of crystalline silicon to examine the effect of temperature and irradiance on the
performance characteristics. The authors of this paper have attempted to develop
a simplified MATLAB/Simulink model that has been validated for performance
characteristics by comparison with commercially available polycrystalline module
from Easy Photovoltaic Private Limited, Ghaziabad, Uttar Pradesh. Then the devel-
oped model has been extended to simulate presently used commercial materials like
Monocrystalline Silicon (Si-mono), Polycrystalline Silicon (Si-poly), Amorphous
Silicon (a-Si) and the emerging like Cadmium Telluride (CdTe), Copper Indium
Selenide (CIS), and Gallium Arsenide (GaAs).
2 PV Module Technologies
Several crystalline silicon based and thin-film PV technologies have been demon-
strated commercially on a large scale in the past few years. In addition, several
emerging PV technologies may be technically and economically competitive in the
future. This subsection briefly describes some of these PV module technologies and
compares their performance, particularly conversion efficiency. Efficiency is a signif-
icant characteristic for comparison of performance of different materials for design
of modules. The efficiency of a PV cell or module is defined as the percentage of the
solar energy striking the cell or module that is converted into electricity [7].
The PV materials considered here can be broadly classified as follows:
20 Simulation of Performance Characteristics of Different … 201
2.1 Crystalline Silicon PV Technologies
Crystalline Silicon (c-Si) technologies that constitute about 80% of the current PV
market have reliable performance with working lifetime of more than 25 years. There
are two types of crystalline silicon PV technologies: monocrystalline (Si-mono or
Sc-Si) and multi-crystalline (Si-poly or mc-Si). The rated DC efficiencies of standard
c-Si PV modules are about 14–16% [7].
2.2 Thin-Film PV Technologies
Thin-film PV cells have a few microns (µm) thick semiconductor layer which is about
100 times thinner than commercial c-Si cells. The most common thin-film semi-
conductor materials include Amorphous Silicon (a-Si), Cadmium Telluride (CdTe),
Gallium Arsenide (GaAs), and alloys of Copper Indium Selenide (CIS). Thin-film
modules have lower DC efficiencies than c-Si modules: about 9–12% for CdTe and
6–9% for a-Si. CdTe-based PV has proficient expressively higher market growth
during the last decade than the other thin-film PV technologies [7].
3 Design of MATLAB-Simulink Model of PV System
The equivalent circuit and mathematical equations used to design MATLAB-

Simulink model of the generalized PV cell have been described in this section.
The model of [1] has been further extended for modeling a module and an array as
described in Sect. 3.2 and simulated as described in Sect. 3.3. It is modified to suit the
modeling of cells of different important materials in Sect. 3.4 to build a generalized
model that is suitable for scaling at all levels of model, i.e., the PV cell, module, and
an array.
The generalized PV model has been developed by using the following nonlinear
voltage-current (V-I) characteristic equation of a solar cell [8]:
q (V +I R )
S V + I RS
I = I P H − I S exp AkTC − 1 − (1)
RS H
where
I PH Photocurrent,
IS Saturation current,
q Charge on an electron, 1.6 × 10−19 C,
k Boltzmann constant = 1.38 × 10−23 J/K,

A Ideality factor,
TC Working temperature,
RS Series resistance,
RSH shunt resistance.
The series loss and the leakage to ground is neglected in this model, i.e., RS = 0
and RSH = ∞.
The photocurrent I PH is obtained by

I P H = I SC + K I TC − TRe f G (2)
where
I SC Short-circuit current.
KI Short-circuit current temperature coefficient.
T Ref Reference temperature.
G Solar insolation.
The saturation current I S is obtained by
⎡ ⎤

3 q EG 1
− 1
TC TRe f TC
IS = I R S exp⎣ ⎦ (3)
TRe f kA
where
I RS Reverse saturation current.

E G Energy bandgap of the semiconductor material used.
The reverse saturation current IRS is represented by
I SC
IRS = (4)
q VOC
exp kTC A
−1
Finally, the output current of the PV cell is given as follows:

q VOC
I = I P H − I S exp −1 (5)
kTC A
These equations form the base for the development of model.

I +
Fig. 1 Ideal model of PV module
3.1 Modeling of PV Module and Array
Since the PV cell produces less than 2 W at 0.5 V approximately, the model with a
particular power rating requires that the number of cells must be connected in series–
parallel configuration. Finally, PV array is a group of several PV modules connected
in series and parallel circuits to generate the required current and voltage.
The Eqs. 4 and 5 can be modified by including these circuits to obtain the output
current of a PV module. Hence, IRS and I with Np parallel cells and Ns series cells
are given as follows:
I SC
IRS = (6)
q VOC
exp N S kTC A
−1

q VOC
I = N P I P H − N P IS exp −1 (7)
N S kTC A
The circuit of an ideal PV module with Np parallel cells and Ns series cells is
shown in Fig. 1 [2].
3.2 Simulation of PV Model
With the help of mathematical equations given in this section, simulation of model
is carried out as explained in the following steps:
Step—1 Model I PH using Eq. 2.

Step—2 Model I S using Eq. 3.
Step—3 Model I RS using Eq. 6.
Step—4 Model I using Eq. 7.
Fig. 2 MATLAB/simulink model of PV module
Step—5 The four models developed are finally interconnected as shown in Fig. 2.
A masked model is designed with dialog box, where the parameters of PV module
can be configured in the same manner in case of standard Simulink block library.
The generalized model was adapted to represent Si-mono PV module, and then
it is simulated at Standard Test Conditions (STC) defined as temperature = 25 °C,
Irradiance = 1 kW/m2 and AM = 1.5, and further validated with respect to a reference
model.
3.3 Simulation of PV Model
The reference module for the simulation is PSS1237, which is commercially available
mono-crystalline module from Easy Photovoltaic Private Limited, Ghaziabad (UP).
The technical parameters specified in manufacturer datasheet of this module are
given in Table 1 [9].
Comparison of proposed model values with reference model values at remarkable
points is summarized in Table 2.
This table depicts the I-V and PV simulation, and experimental results show a good
agreement in terms of current at maximum point, voltage at maximum point, and
maximum power. Hence, the designed model has been validated with a reasonable
accuracy. However, Table 2 shows that the modeling error in power lies in the range
Table 1 Technical parameter specifications of PSS1237PV module

Characteristics Specifications
Nominal voltage, V(volts) 12
Open circuit voltage, Voc (volts) 21
Voltage at maximum power, Vmax (volts) 16.8
Short circuit current, Isc (Amps) 2.55
Current at maximum power, Imax (Amps) 2.2
Maximum rated power, Pmax (watts) 37
No. of cells/module 36
The electrical specifications are under standard test conditions of irradiance of 1 kW/m2 , cell
temperature of 25 °C, and AM of 1.5
Permissible range of error is ± 3%
Table 2 Comparison of
Remarkable points Reference model Proposed model
proposed model values with
value value
reference model values
(temperature = 25 °C) and Imax 2.2 2.55
irradiation G = 1 kW/m2 ) Vmax 16.8 22
Pmax 37 45.24
5–9%. The effect of series and shunt resistance has not been taken into account in
the model for the sake of simplicity, i.e., Rs = 0 and Rsh = ∞. Hence, the error that
has been introduced on this account is more than ± 3% which is admissible as per
the limit specified on the manufacturer’s datasheet.
3.4 Model for Comparative Study
The masked model developed is now available in the form of user-friendly interface
that follows the style of standard MATLAB/Simulink block library components, and
an attempt has been made in this paper to use the developed model for conducting
simulative studies by considering the effect of change of material as shown in Fig. 3.
This model is adapted by using suitable parameters in order to represent the following
materials, which are commonly used for fabrication of PV modules:
• Monocrystalline Silicon (Si-mono)
• Polycrystalline Silicon (Si-poly)
• Amorphous Silicon (a-Si)
• Cadmium Telluride (CdTe)
• Copper Indium Selenide (CIS)
• Gallium Arsenide (GaAs).
Fig. 3 Final model
The results of simulation for these different materials have been presented in the
next section.
4 Simulation of Model for Different PV Materials
Six different models based on commonly used PV materials were simulated by

incorporating their relevant parameters in the proposed model. These are as follows:
• Band-gap energy and
• Ideality factor.
The ideality factor and bang-gap energy are dependent on PV technology [9] and
is listed in Table 3.
These six models hence developed were simulated at STC. The results are
tabulated in Table 4.
Table 3 Ideality factor (A)

Technology Ideality factor Energy bandgap
and energy bandgap EG (in
eV at 300 k) for PV Si-mono 1.2 1.1
technology [2] Si-Poly 1.3 1.1
a-Si 1.8 1.7
CdTe 1.5 1.49
CIS 1.5 2.42
GaAs 1.3 1.43
Table 4 Comparison of different PV modules at STC (temperature = 250 C) and irradiation G =

1 kW/m2 )
Name of material Imax Vmax Pmax Fill factor %Efficiency
(Pmax/Voc*Isc) (Pmax/G*Ac)*100)
Si-mono (sc-Si) 2.55 22 45.24 0.8448 11.51
Si-poly (mc-Si) 2.55 23.5 40.16 0.7947 10.82
a-Si 2.54 23.5 38.76 0.7499 10.21
CdTe 2.52 28 35.83 0.669 9.12
CIS 2.52 28 35.83 0.669 9.12
GaAs 2.55 23.5 40.16 0.7947 10.82
Table 4 shows that the efficiency of Si-mono is the highest among all the materials
at standard test conditions closely followed by Si-poly. It can also be seen that the
materials having same ideality factor having same range of output current, voltage,
and power like Si-poly/GaAs and CdTe/CIS.
Although the development of new materials like CdTe, CIS, and GaAs has been
received by the academic community with large fan force, they have not received
favor of practical venture capitalist. Hence, Si-mono has emerged as the best choice
among the six studied materials. The added advantage is that its cost is lowest among
these materials and is available in abundance, which further justifies its widespread
popularity.
5 Conclusion
A MATLAB/Simulink PV model has been developed for comparing the perfor-

mance of six different PV materials in this paper. The proposed model was validated
with PSS1237. The simulation results revealed a good agreement with the practical
data from the manufacturer, Easy Photovoltaic Private Limited, Ghaziabad (Uttar
Pradesh). The manufacturer’s datasheet specifies that the rated power may vary by
±3%. However, Table 2 shows that the modeling error in power is somewhat higher
than that of specified limit. The influence of series and shunt resistance has not been
taken into account for the sake of simplicity, i.e., Rs = 0 and Rsh = ∞. Hence, the
error that has been introduced on this account is more than ±3% admissible as limit
specified on the manufacturer’s datasheet. The validated model is used for carrying
out study of PV modules manufactured from six commonly used PV materials. These
were simulated by incorporating their relevant parameters in the proposed model.
Their simulation for studying the behavior of different materials has depicted that
the Monocrystalline Silicon shows the highest efficiency among all the materials at
standard test conditions. It has also been found that all the materials having the same
ideality factor exhibit similar behavior with respect to temperature and irradiation.
Finally, it can be concluded that Si-mono is the best PV material for commercial
application.
References
1. Khyani HK, Vajpai J (2013) Mathematical modeling and experimental validation of performance
characteristics of solar photovoltaic modules. Int J Appl Innov Eng Manag (IJAIEM) 2(11):295–
301. ISSN 2319–4847
2. Tsai HL, Tu CS, Su YJ (2008) Development of generalized photovoltaic model using
MATLAB/simulink. In: Proceedings of the World congress on engineering and computer science
(WCECS ‘08), San Francisco (USA). https://www.iaeng.org/publication/WCECS2008/WCE
CS2008_pp846-851.pdf
3. Longatt FMG (2005) Model of photovoltaic module in matlab. In: 2nd international conference
on Iberoamerican congress to electrical engineering students, electronics and computing (II
CIBELEC: 2005), pp 1–5
4. Bhatt HG, Thakker RA (2011) Matlab based simulation of photovoltaic solar cell and its array
at different temperature values. In: National conference on recent trends in engineering &
technology, B.V.M. Engineering College, Gujarat, pp 1–4
5. Mahmoud Y, Xiao W, Zeineldin HH (2012) A simple approach to modeling and simulation of
photovoltaic modules. IEEE Trans Sustain Energy 3(1):185–186
6. Panwar S, Saini RP (2012) Development and simulation of solar photovoltaic model using
MATLAB/simulink and its parameter extraction. In: International conference on computing and
control engineering (ICCCE’2012)
7. SunShot Vision Study (2012) A technical report on solar PV technology, Prepared by US
Department of Energy. https://www1.eere.energy.gov/solar/pdfs/47927.pdf
8. Khyani HK (2013) Modeling and simulation of solar photovoltaic systems. M.E. thesis, Elec-
trical Engineering Department, Faculty of Engineering (M.B.M. Engineering College) JNV
University, Jodhpur, pp 27–75
9. Solar PV Module (2019). https://www.easyphotovoltech.com/solar-module.html,last. Accessed
12 Oct 2019
Chapter 21
Stabilization Analysis of Clay Soil
by Using Waste Tile Particles
Saraswati Chand Dhariwal and Rajat Mangal
1 Introduction
Far reaching soils present huge geotechnical and auxiliary building difficulties the
world over, with costs related with extensive conduct evaluated to run into a few
billion yearly. Broad soils will be soils that experience critical volume change related
with changes in water substance. These volume changes can either as swell or in
the structure shrinkage and this is the reason, they are at some point known as
swell/shrivel soils. Key points that need ID while overseeing expansive soils include
soil properties, suction/water conditions, water content assortments transient and
spatial, for instance created by trees, and the geometry/firmness of foundations and
associated structures. So as to make an improved soil material which has the ideal
designing properties, it is basic for the dirt to experienced adjustment for at least one
properties. Either by doing the modification by mechanical or substance, this proce-
dure is called as soil adjustment. The purpose of soil stabilization not only limited to
enhance the load bearing of the soil capacity but also improve the shear strength, filter,
drainage system, permeability, enhance soil resistance to the weathering process and
traffic usage to meet specific engineering projects requirement [2, 6].
S. C. Dhariwal (B)
M.B.M. Engineering College, J.N.V. University, Jodhpur, Rajasthan, India
e-mail: sonudhariwal14@gmail.com
R. Mangal
J.I.E.T Engineering College, Jodhpur, Rajasthan, India
e-mail: rajat.mangal0@gmail.com

https://doi.org/10.1007/978-981-15-8820-4_21
210 S. C. Dhariwal and R. Mangal
2 Scope of Present Work
As a huge wastage of tiles waste is available in world so it can be used as admixture

for stabilization of clay soil there is a great scope of stabilization of clayey soil with
the admixture of tiles particles for construction of pavements, airfields and helipads.
the aim of present work is the beneficial and economical utilization of such wastages
for improving properties of clayey soil which is to be used as a base material for
streets development and appropriate establishment material for different sorts of
super structures [3].
Utilization of tiles particles for improvement of a soil is a sustainable and cost-
effective technique. Since tiles particles can be easily available hence it was thought
to be utilized as an admixture with which clay soil can be utilized as a low-cost
building material. On the other hand, the problem of the disposal of tiles particles
can be overcome by using it for stabilization of clay soil [1].
3 Materials Used in This Study
3.1 Clay Soil
Clay soils contain minerals that are capable of absorbing water. When water is added
in the clay soils, they increase in volume. And again, more water added, the more their
volume increases of the clay soils. this changing in volume can exerts enough forces
on a building or other structure this cause damage the structure. Broken establish-
ments, floors, and storm cellar dividers are common kinds of harm done by expanding
soils. Clay soils will also be shrinking when they remove the water content. The clay
sample for in this study were collected from Mundwa city in Mundwa Tehsil in
Nagaur District of Rajasthan State, India. It belongs to Ajmer Division. It is found
31 km towards South from District base camp Nagaur [4].
3.2 Waste Tile Particle
A waste ceramic tile particle is an inorganic, non-metallic solid prepared by the action
of heat and subsequent cooling. Waste ceramic tile material may have crystalline or
partly crystalline structure because most common ceramics are crystalline materials.
Ceramic Tile waste is cheap and non-reusable material, it is shown in all construction
area and easy to collect. By the using of ceramic Waste Tile Particles to reduce the
waste materials in earth and economical. The Waste ceramic tiles used in this study
work were of hindware company. The waste ceramic tiles were bought from the local
tiles shop from Jodhpur city, Rajasthan (India) [5, 7] (Fig. 1).
21 Stabilization Analysis of Clay Soil by Using … 211
Fig. 1 Tiles particles
4 Experimental Programme
The Following Tests Were Conducted to Determine the Engineering Properties of

The Clay Soils Mixed with Tiles Particles:
• Light compaction Test (Standard Proctor Test) for determining different dry
densities of above-mentioned clayey samples and optimum moisture content of
them.
• CBR test to determine % CBR values for clayey mix composition with wastes
tile particles.
• Direct shear test to determine shear strength of clayey mix composition with waste
tile particles (Figs. 2 and 3).
Fig. 2 CBR test mix composition in mould
Fig. 3 Direct shear test apparatus

Table 1 Dry density

S. No. % water added (by weight) Dry density (gm/cc)
variation with water content
for clay soil 1 8 1.76
2 10 1.78
3 12 1.79
4 14 1.81
5 16 1.83
6 18 1.74
7 20 1.71
8 22 1.69
9 24 1.65
10 26 1.64
Maximum Dry Density

DRY DENSITY (gm/cc)
1.85
8
1.8
1.75 10
1.7
12
1.65
1.6 14
1.55
16
1.5
8 10 12 14 16 18 20 22 24 26 18
WATER CONTENT (%) 20
Fig. 4 Dry density variation of clay soil with water content
5 Test Results
5.1 Standard Proctor Compaction Test
The maximum dry density (MDD) of the clay soils without any admixture mixed is
obtained as 1.83 gm/cc at the optimum moisture content (OMC) equal to 16%. The
dry density variation with water content for clay soil is tabulated below in Table 1
and graphically shown below in Fig. 1 (Fig. 4).
5.2 California Bearing Ratio Test
The results of the unsoaked CBR tests conducted at MDD of 1.83, 1.78 and 1.69
gm/cc in Tables 2, 3 and 4.
The graphical variation of the percentage CBR values in unsoaked conditions for
different mix compositions of admixture with Clay Soil is shown in Fig. 5.
Table 2 Mix compositions, symbols for unsoaked CBR test at MDD 1.83 gm/cc and % CBR value
of clay soil with each mix composition
S. No. Mix composition Symbol % CBR value
1 0.05% Tiles Particles + Clay CB1 7.153
Percentage (%) CBR Value Variation in Mix Compositions in

Unsoaked Conditions
10
% CBR VALUE
0.05
0.075
5
0.25
0.5
0 0.75
1.83 1.78 1.69 1
PERCENTAGE OF TILES PARTICLES
Fig. 5 Variation of % CBR values in unsoaked conditions for different mix compositions of
admixture with Clay Soil of dry density 1.83, 1.78 and 1.69 gm/cc
From the above graph it is clear that among the three samples taken for the study,
the maximum CBR value is obtained for dry density of Clay Soil of 1.69 gm/cc. It
is also stated here that the maximum CBR value is obtained when the percentage by
weight of admixture (Tiles Particles) mixed with the Clay Soil is 0.25%.
5.3 Direct Shear Test
The results of Direct Shear test at MDD 1.83 gm/cc of Clayey Soil mix with
percentage of waste tiles particles are shown in Tables 5, 6 and 7 (Figs. 6, 7 and 8).
Table 5 Variation of ϕ with percentage of tiles particles at MDD 1.83 gm/cc

S. No. Mix composition (Degree)
1 DB1 14.50°
2 DB2 17.76°
3 DB3 16.13°
4 DB4 14.40°
5 DB5 13.82°
6 DB6 15.47°
Table 6 Variation of C with percentage of tiles particles at MDD 1.83 gm/cc

S. No. Mix composition C (Cohesion)
1 DB1 0.338
2 DB2 0.345
3 DB3 0.369
4 DB4 0.348
5 DB5 0.387
6 DB6 0.408
Table 7 Variation of shear stress with normal stress for all mix composition at MDD 1.83 gm/cc
Shear Stress (kg/cm2 ) for each mix Normal stress (kg/cm2 )
composition at MDD 1.66 gm/cc 0.5 1.0 1.5
0.05% Tiles Particles + Clay (DB1) 0.357 0.476 0.614
Variation of φ with Percentage of Tiles

Particles
20 17.76
FRICTION ANGLE
16.13 15.47
14.5 14.4 13.82
15 0.05
10 0.075
0.25
5
0.5
0 0.75
0.05 0.075 0.25 0.5 0.75 1
1
PERCENTAGE OF TILES PARTICLES (%)
Fig. 6 Variation of ϕ with percentage of tiles particles at MDD 1.83 gm/cc
Variation of C (Cohesion) with Percentage of Tiles

Particles
0.5
0.387 0.408
0.369
COHESION (C)
0.4 0.338 0.345 0.348 0.05

0.3 0.075
0.2 0.25
0.5
0.1
0.75
0
0.05 0.075 0.25 0.5 0.75 1 1
PERCENTAGE OF TILES PARTICLES (%)
Fig. 7 Variation of C with percentage of tiles particles at MDD 1.83 gm/cc
Variation of Shear Stress with Normal Stress for All

Mix Composition
SHEAR STRESS (kg/cm2)
0.8
0.7
0.6
0.5
0.4 0.5
0.3 1
0.2
0.1 1.5
0
0.05 0.075 0.25 0.5 0.75 1
NORMAL STRESS (kg/cm2)
Fig. 8 Variation of shear stress with normal stress for all mix composition at MDD 1.83 gm/cc
6 Conclusions
The following conclusions can be made from the above study:

• It is observed that the Maximum Dry Density (MDD) of the clay soil is first
Increased and after that it will be Decreased. In this study the Maximum Dry
Density is Obtained 1.83gm/cc at the Optimum Moisture Content (OMC) to 16%.
• It is observed that the CBR values first increases when the percentage by weight
of admixture (Tiles Particles) mixed with the clay soil increases up to 0.25%
and then with the increase in percentage by weight of admixture the CBR values
decreases.
• The above observation is same for all the three samples and thus it is concluded
that the variation of CBR values with the percentage by weight of admixture is
independent of the dry density of the clay soil.
• It is also seen here that the maximum CBR value is obtained when the percentage
by weight of admixture mixed with the clay soil is 0.25% irrespective of the dry
density of the clay soil.
• Among the three samples taken for the study, the maximum CBR value is obtained
for dry density of clay soil equal to 1.69 gm/cc.
• It is observed that the maximum angle of friction is 17.76° and cohesion (C) is
0.408 is determined. And the Maximum shear stress value is obtained at 1.5 kg/cm2
of normal stress in the Direct Shear Stress Test.
• Tiles Particles can be used as admixture and thus its recycling helps in reducing
environmental problem related to its dumping and/or disposal.
References
1. Arora KR (2016) Soil mechanics and foundation engineering (geotechnical engineering)

2. Bhuvaneshwari S, Robinson RG, Gandhi SR (2005) Stabilization of expansive soils using fly
ash. In: Fly ash utilization programme, FAUP, TIFAC, DST, vol 8, pp 5.1–5.9
3. Cokca E (2001) Use of class C fly ashes for the stabilization—of an expansive soil. J Geotech
Geo Environ Eng 127:568–573
4. Purohit DGM, Ametha NK, Wayal AS (2008) Characteristics, problems and remedies of
expansive soils of Rajasthan, India. In: E.J.G.E, vol 13
5. Robert M, Brooks (2009) Soil stabilization with fly debris and rice husk ash. Int J Res Rev Appl
Sci 1:209–217
6. Singh A (2009) Basic soil mechanics and foundation. CBS Publishers and distributors, India
7. Tiwari S, Tiwari N (2016) Soil stabilization using waste fiber materials. Int J Innov Technol Res
4(3):2927–2930
Chapter 22
Renewable Energy as Biofuel
from Mirabilis Jalapa Seed Oil
Akleshwar Mathur and Harish Kumar Khyani
1 Introduction
The improvement in the living standards of a citizen is directly or indirectly related

to the energy as a critical input toward the same. This fact is now realized by the
Government so it has become one of the most important ambitious announcements.
All villages should get electricity at 24 h for every day throughout the year by 2019,
The 175 GW of renewable energy capacity by the year 2022 followed by reduction in
intensity of energy emissions by 33–35% by the year 2030 with increase in sharing
of non-fossil fuel-based capacity in the electricity blend at above 40% in the same
year [1].
Fossil fuels are more commonly used as source of energy in developing countries
like India. Since the natural energy resources are continuously harnessed and demand
is increasing the price of fossil fuel are very high, moreover, these fuel induces climate
change, environmental pollution and greenhouse effect so there is an urgent need to
explore new renewable energy sources to overcome these problems. Biofuels are
considered as a promising source of energy as the best natural substitute for diesel.
The present paper belongs to the recent status of non-conventional energy sources
in our country with special emphasis on biodiesel development, central and state
government policies and the physicochemical properties of selected species [2].
The good quality of biofuels can be blended easily with gasoline and can be
used as an individual fuel. The biofuels emit lesser amounts of polluted gases so
are carbon neutral and safe as compared with conventional petroleum fuels. The
estimated production of biofuel in India was 500–600 million liters in 2017–18.
A. Mathur (B) · H. K. Khyani

JIET, Jodhpur, Rajasthan, India
e-mail: akleshwar.mathur@jietjodhpur.ac.in
H. K. Khyani
e-mail: harish.khyani@jietjodhpur.ac.in

https://doi.org/10.1007/978-981-15-8820-4_22
220 A. Mathur and H. K. Khyani
India can fulfill their fuel demands totally by biofuels. Our country produces a large
amount of cellulosic material, which can be converted into biofuels [3, 4].
1.1 Indian Energy Scenario
On 15 August 2015, Dr A. P. J. Abdul Kalam, President of India, quoted a remark-

able commitment that, “By the year 2030, India should achieve energy Indepen-
dence through solar power and other forms of renewable energy.” This would be
like to achievable through the fast-growing applications of new and novel energy
resources. The targets of renewable energy sources till 2021–22, Indian Energy
Scenario fuel wise and contribution of various renewable sources in total energy
mix are summarized in Tables 1, 2 and 3, respectively (Tables 4 and 5).
Although biofuels have a small fraction as compared with other resources,
Biofuels are being considered as important commercial fuels, resulting in tremen-
dous growth of biodiesel industry. By observing the current fuel-wise sharing, 2.7%
of total fuel is being sourced from biofuels [5–7].
Table 1 Target of different

S.No Source Target ( GW)
renewable sources by MNRE
by the year 2021–22 [1] 1 Solar 100
2 Wind 60
3 Biopower 10
4 Small hydropower 5
Total 175
Table 2 Indian energy

Fuel MW % share
scenario fuel wise as on
30.09.2019 Coal 196,894.50 54
Lignite 6260.00 1.7
Gas 24,937.22 6.8
Diesel 509.71 0.14
Hydro 45,399.22 12.5
Nuclear 6,780.00 1.86
RES** (MNRE) 82,588.95 23
Total 3,63,369.59
22 Renewable Energy as Biofuel from Mirabilis … 221
Table 3 Contribution of various renewable sources in total energy share as on 30.09.2019

Fuel MW % share w.r.t. to RES (82,588.95) % share w.r.t. to Overall
(3,63,369.59)
SHP 4610.81 5.6 1.3
Wind 36,930.32 44.7 10.3
Solar 31,101.71 37.6 8.7
Biopower 9946.11 12.1 2.7
Total 82,588.95
Table 4 Physicochemical properties of seeds and oil

Sr. no Property of seeds/oil Value
1 Moisture content 4.0%
2 Oil content 3.5%
3 Refractive Index at room temperature 1.466
4 Specific gravity 0.789
5 Acid value (mg NaOH/g of oil) 1.345
6 Saponification value (mg KOH/g of oil) 174
7 Iodine value (g I2 /100 g of oil) 83
8 Total saturated fatty acids 12.57%
9 Total unsaturated fatty acids 87.43%
10 Poly unsaturated fatty acids 94.39%
11 Category of oil Non-drying
Table 5 Comparison of mirabilis jalapa seed oil with jatropha and karanj seed oils [16, 17]
Sr. Component Jatropha curcus Pongammia pinneta Mirabilis jalapa
1 Oil Content 35 38 3.5
2 Specific Gravity 0.918 0.934 0.789
3 Saponification Value 195 192 174
4 Iodine value 86 90 83
5 % Unsap matter 1 1 1
6 Myristic acid 0.6 NIL NIL
7 Palmitic acid 14 5.5 3.41
8 Stearic acid 7.5 5.6 1.01
9 Oleic acid 49.9 56 1.20
10 Linolenic acid – 2.6 60.70
11 Linoleic acid 28 14 33.69
12 Unusual component Absent Absent Absent
13 PUFA 28.0 16.8 94.39
1.2 Biofuels in India
The gasoline blend rate in 2016 was remarkable as 3.3%, which results the production
of 143 billion liters biofuels with an average growth of biofuels as 2.5% per annum
in the next year [5]. The USA, Brazil and Germany are the first three chief producers
of biodiesel in the world, and the 43.5% of global biofuel production is achieved
by the USA [6]. As earlier reported that the use of biofuels results reduction in the
dependence of the petroleum products, control on CO2 , SO2 , NO2 -like greenhouse
gas emissions and other major problems, with additional advantages. India is a well-
recognized biodiversity ecosystem and produces tones of biowaste from plants, so
this resource becomes more productive compared with other countries. Sugarcane
and sugar mother liquor, seed-producing crops like corn, soybean, sorghum and
non-edible oils jatropha, Pongammia, wild sunflower verities, cellulosic materials,
are considered as raw materials for biofuels [7, 8]. The current initiatives of the
Government including the Make in India and Swatch Bharat Abhiyan and various
offer great opportunity to catch the targets of increasing Farmer’s Income, Import
Reduction, Employment Generation, Waste to Wealth production, etc.
The utilization of dry and wastelands for the cultivation of wild or non-edible
seed verities would result to generate wealth for the local population with more
important effect in the sustainable development. The goal of the policy is to increase
the blending percentage to provide biofuels in the market at low cost. According
to the recent literature, ethanol blending percentage in petrol is around 2.0% and
biodiesel blending percentage in diesel is less than 0.1%. Blending of ethanol in
petrol to 20% and blending of biodiesel in diesel is 5% proposed by the year 2030.
This target is achievable as India is the seventh largest ethanol-producing country in
the world. In addition to this, the Indian government aims to reduce its 10% of crude
oil imports by 2022 [9, 10].
1.3 Biodiesel Blending Program
Feedstock availability for biodiesel is a key factor related to the growth of this
sector. The overall blending percentage of biodiesel in diesel is less than 0.5% and
the biodiesel is manufactured from imported sources. Thus to ensure domestic raw
material source is integral for long-term success of this program.
In addition to the raw materials, household resources also have great importance,
used or waste cooking oil termed as UCO and WCO, respectively, offers a potential
source of biodiesel production. The variable and cultural food products provide a
rich amount of waste oils, which are not safe for reuse but are reused through various
small eateries, vendors and traders resulting health issues. The entry of UCO in food
stream and developing a suitable collection mechanism to augment its supply for
biodiesel production could be run simultaneously.
1.4 Initiatives by the Government of Rajasthan
The biofuel rules 2019 were released on the eve of the World Biofuel Day, August
9, 2019, by the Rajasthan government and Rajasthan is the first state in the country
to bring the policy for biofuel after the notification of the Centre [11].
1.5 Requirement of New Sources and Current Species
Although biodiesel has a number of advantages, the selection of raw material is a

challenge to the producers so production of a cost-effective fuel is not achieved as
edible oils are being used in biodiesel production. Already the use of edible oils
in industries has resulted in the high cost, so one would not capable to increase
again the consumption of edible oils in a non-edible sector. To make economical
biodiesel, non-edible, low cost, easily available and low-grade oils with value-added
by-products to be searched as alternative raw material. Figure 1 indicates the data
about the Production and Consumption of Ethanol in India of the last 10 years and
2, the import of crude oil in India in the same time period (Fig. 2).
The demand and supply from other countries expanding huge amount of finance
are not good for the economy of a developing country, especially India as a number
of plant resources are still unexploited by us. So it the correct time to explore the
new and novel resources of oils for the fuel industry.
In the above-mentioned scenario of oils, the phytochemical analysis of number of
seed oils has been in progress since the last two decades. Mirabilis jalapa commonly
known as 4 o’clock flower or the marvel of Peru is a member of botanical family
Nyctaginaceae. It is a non-conventional, inexpensive, low-grade seed oil and unex-
ploited for biodiesel production. Furthermore, the crude oil is rich in a variety of
phytochemicals like flavonoids, triterpenes, proteins, steroids and alkaloids with
Mirabilis antiviral proteins abbreviated as MAPs. Therefore, use of mirabilis jalapa
Fig. 1 Production and consumption of ethanol in India

Fig. 2 Import of crude oil in India
seed oil as raw material for biodiesel production would not only make the process
economical but also generate value-added phytochemicals or bioactive compounds.
In the present investigation, physicochemical properties of this seed oil have been
reported with fatty acid composition [12, 13].
As per literature available, this oil has not been a viable biodiesel feedstock due
to its low oil content and less emphasis on fixed oils as compared with corn oil,
canola oil, rice bran oil, etc., but as earlier discussed, it would be a good source of
raw material [14].
2 Materials and Methods
The seeds were collected from cultivated plants after maturation by hand-picking
methods. The dust were removed by washing seeds with water. Seeds were air
dried at 60 °C for 7 h in oven. The seeds were crushed by Mortar and pestle into
a coarse powder. Petroleum ether of 40–60° boiling point range was selected as
extraction solvent for oil using Soxhlet apparatus. The solvent was recovered by
Rota-evaporator under vacuum. Percentage of moisture, oil percentage and protein
content of the seeds were determined recommended by American Oil Chemical
Society as shown in Table 1 [15–19].
The analytical TLC was prepared by coating 0.25 mm layers of Silica gel G using
methanol solvent on glass plates. A mixture of petroleum ether and diethyl ether in
80:20 ratio was applied as eluting solvent. The spots were visualized by exposure to
Iodine vapor and then charring with Sulfuric dichromate spray.
The refractive index was recorded on Abbey Refractometer. GLC analyses were
carried out with Varian Vista 6000 Gas Liquid Chromatograph using DEGS and SE-
30 columns. Various TLC methods were used for further confirmation. The Refractive
Index was determined by the Adobe Refractometer, at 30 °C. Density bottle was used
for specific gravity. ISO 3657 (1988) indicator method was applied to determine the
saponification value [20, 21]. Acid value was determined by as FFA/2.ISO 3961
(1989) method was used to determine the Iodine value. The GLC of methyl esters
was done to analyses the fatty acid composition of the seed oil as usual methods
[15–17].
The oil content in the collected sample was found as 3.5% which may be due to
environmental and climatic conditions and also affected by the mode of extraction.
In earlier reports, it has been found in the range of 3–5%. The oil is light, transparent,
odorless, colorless and clear. The refractive index of oil was found as 1.466, specific
gravity as 0.72, acid value as 1.345 (mg NaOH/g of oil), saponification values (mg
KOH/g of oil) as 172. The Iodine value (g I2/100 g of oil) was found as 83. The oil
is categorized as a non-drying oil. The physicochemical properties of seeds and oils
determined are summarized in the table.
Total Saturated Fatty Acids: The total amount of saturated fatty acids was found
as 12.57%. The fatty acids like Myristic, Palmitic and Behinic acids were absent in
selected species.
Total Unsaturated Fatty Acids: The total amount of oleic acid, linoleic and
linolenic acids was found as 87.43% as the major component of all the fatty acids.
Other unsaturated acids were also found in small amount.
Poly Unsaturated Fatty Acids (PUFA): The combined content of linolic and
linolenic acid was found as 94.39%.
Category of Oil: On the basis of PUFA content, the seed oil of Mirabilis jalapa
is categorized as non-drying oil.
3.1 Comparison with Earlier Studies and Other Seed Oils
The comparison of present findings with earlier reports was done [22, 23]. The
richness of Poly Unsaturated Fatty Acids was found similar except the presence of
oxygenated fatty acid component due to environmental and species variation factors.
The table indicates the comparative chemical composition of seed oils of Jatropha
curcus and Pongammia pinneta seed oils with mirabilis Jalapa seed oil. These two
plant species are commonly in practice for biodiesel preparation. It is clear that most
of the properties resemble with jatropha and karanj oil.
4 Conclusion and Future Scope
The physicochemical properties of the mirabilis jalapa seed oil seem to have good
scope and applications in industrial approaches. On the basis of unsaturated fatty
acids, it is a non-drying oil so it can be used as an additive in biodiesel and lubricants
also in manufacturing of medicated cosmetics, drug binder, etc. Further the yield of
the oil is influenced by environmental conditions, variety and age of plant, collection
timings, storage, etc. On the basis of the present results and previous research find-
ings, more applications could be suggested. It is concluded that Mirabilis jalapa oil
could be a substitute for preexisting seed oil. It can be cultivated as a medicinal crop-
producing specific alkaloid. Since the oil is found in small amount hence it might be a
part of research regarding oil improvement experiments to prove a promising source
of Indian economy, especially in our desert zone. It is suggested that the cultivation
of this plant should be granted under government projects.
References
1. http://www.cea.nic.inhttp://cea.nic.in/monthlyinstalledcapacity.html. Accessed 11 Nov 2019

2. Order no 30/80/2014–15/NSM, dated 1st July 2015
3. Ho DP, Ngo HH, Guo W (2014) A mini review on biofuel. Bio resource technology, pp 1–31
4. Kitani O (1999) CIGR book of agriculture engineering, energy and biomass engineering, vol
V. ISBN 0-929355-97-0
5. Moon, Pakrasi HB (2013) Front Microbiol 4:1–14. Article 246
6. Genetically engineering cyanobacteria to convert CO, water and light into the long-chain
hydrocarbon farnesene, Bioenergy and Biofuels. Appl Microbiol Biotechnol 98:9869–9877
(2014)
7. Alternative fuels data center—fuel properties comparison. www.afdc.energy.gov
8. Berla BM, Saha R, Immethun CM, Maranas CD, Seok T, Synthetic biology of cyanobacteria:
unique challenges and opportunities
9. The Global Dynamics of Biofuels (2007) Brazil institute Special report (3):1–8
10. Ethanol industry outlook (2012) Renewable fuel assosiation America
11. https://www.biofuel.rajasthan.gov.in/about.aspx
12. Chisti Y (2008) Biodiesel from microalgae beats bioethanol. Trends Biotechnol 26(3):126–131
13. Miller CD, Bazore K, Bartow M (1955) Fruits of Ha-waii, 2nd edn. University of Hawaii Press,
Hawaii
14. Barrett OW (1928) The tropical crops. The Macmillan Co., New York
15. Kennard W, Winters H (1960) Some fruits and nuts for the tropics. Misc. Pub. 801, Agricultural
Research Service, US Dept. Agric, Washington
16. Deora MA, Jaiswal P, Mathur A, Sherwani MRK (2003) Fatty acid composition of some minor
seed oils from arid zone of Rajasthan. JICS 80:141–142
17. Jaiswal P, Mathur A, Sherwani, Phytochemical studies of some seed oils from arid zone of
Rajasthan, Oriental J Chem 19(2):487–488
18. Ahmad MS, Rauf A, Mustafa J, Sheikh MO (1984) An 8-hydroxyoctadeca-cis-11, 14-dienoic
acid from Mirabilis Jalapa seed oil. Phytochem 23:2247–2249
19. Kyriakidis NB, Katsiloulis T (2000) Calculation of iodine value from measurements of fatty
acid methyl esters of some oils: comparison with the relevant American oil chemists society
method. J Am Oil Chem Soc 77(12):1235–1238
20. Standard methods for the analysis of oils, fats and derivatives 1st Supplement to the 7th Edition,
International union of pure and applied chemistry commission on oils, fats and derivatives.
21. Endo Y (2013) Analytical methods to evaluate the quality of edible fats and oils: the JOCS
standard methods for analysis of fats, oils and related materials; (2018) Advanced methods. J
Oleo Sci 67(1):1–10
22. Mathur A (2017) A comparative physicochemical analysis of crude and refined jamun
(Syzygium cumini) seed oil. Iconic Res Eng J 1(3)
23. Mathur A (2014) Renewable energy sources from Moringa Oleifera seed oil: a rich source of
oil for bio diesel. Int J Comput Appl (0975–8887) National conference on innovations and
recent trends in engineering and technology (NCIRET-2014)
Chapter 23
A Comparative Study of Various
Approaches to Lossy Image Compression
Process
Nitesh Agarwal and Rajendra Purohit
1 Introduction
1.1 Image
An image sensing device captures a real-world image and converts it into a digital
image, which is represented in a two-dimensional (2D) discrete space and denoted
by f (x, y). The dimension of digital image f (x, y) is represented by N rows and M
columns in discrete space, where N, M are integer values and f (x, y) represents the
gray value at location (x, y). x, y are varied from 0 to N-1 and 0 to M-1, respectively
[1] (Fig. 1).
1.2 Image Compression Model
Image compression model consists of two separate modules named as encoder and
decoder as illustrated in Fig. 2. The encoder uses some data compression algorithm
to compress input image, whereas decoder may be viewed as a reverse procedure
of encoder to decompress the output of the encoder. In image compression, decom-
pressed image can be or cannot be an exact replica of the input image; if decompressed
image is an exact replica of input image then it is lossless model, else it is a lossy
model. If the input image of resolution M × N is represented by f (x, y) and decom-

pressed image is represented by f (x, y), then for lossless Image Compression can
N. Agarwal (B) · R. Purohit

Jodhpur Institute of Engineering and Technology, Jodhpur, India
e-mail: niteshagarwal.234@rediffmail.com
R. Purohit
e-mail: Rajendra.purohit@jietjodhpur.ac.in
https://doi.org/10.1007/978-981-15-8820-4_23
228 N. Agarwal and R. Purohit
Fig. 1 Digitization of a continuous image [1]
Mapper Quantizer Symbol Coder
Encoder
Symbol Decoder Inverse Mapper
Decoder
Fig. 2 Image compression model
be given as
f (x, y) = fˆ(x, y), ∀x ∈ [0, M − 1], ∀y ∈ [0, N − 1] (1)
and for lossy Image Compression as
f (x, y) = fˆ(x, y), f or any(x ∈ [0, M − 1], y ∈ [0, N − 1]) (2)
2 Methodology
This research is based on lossy compression and embeds the n-Modulus Method in
lossy compression and analyzes it with three different mathematical transformations,
DCT, DST, DHT (Discrete Hartley Transform), and two different quantization tables.
23 A Comparative Study of Various Approaches … 229
Fig. 3 Functional block diagram of proposed work
A functional block diagram of this research is shown in Fig. 3. Proposed process

has three phases: first is compression, second is decompression, and the third is
analysis.
2.1 Compression and Decompression
Compression starts with dividing the input image into blocks of size 8 × 8 each.
The total number of blocks depends upon the resolution of input image, if the input
image has resolution M × N , then the total number of blocks represented by p can
be defined as shown in (Fig. 4)
M×N
p= (3)
64
After the decomposition of image into blocks, each block is transformed into
another domain (frequency domain) using three basic mathematical transformation
8*8 8*8
block 1 block 2
8*8 8*8
block 3 block 4
Input Image of size Blocks of size 8*8

16*16
Fig. 4 Division of image into blocks

function DCT, DST, and DHT. Each transformed block then quantizes using standard
quantization table and generates quantized blocks. Each quantized block is then
operated by the n-modulus method and, finally, RLE performs the final compression
phase and generates compressed files of input image [2, 3].
2.2 Transformation
It is based on a variety of different mathematical transformations. The choice of

specific transformation depends on the amount of transformation error that can be
tolerable and availability of computational resources.
A transformed block can be represented by T (u, v) for input image f (x, y) of
resolution M X N and can be expressed as

M−1 N −1
T (u, v) = f (x, y)r (x, y, u, v), (4)
x=0 y=0
where r (x, y, u, v) is termed as the forward transformation kernel and its inverse can
be defined as

M−1 N −1
f (x, y) = T (u, v)s(x, y, u, v), (5)
u=0 v=0
where s(x, y, v) is termed as the inverse transformation kernel. The present work is
processed by three basic transformation functions individually, named as DCT, DST,
and DHT.
2.2.1 Discrete Cosine Transform (DCT)
DCT has the same transformed kernel in compression and decompression phases
and can be represented as follows:

π (2x + 1)u π (2y + 1)v
r (x, y, u, v) = s(x, y, u, v) = α(u)α(v) cos cos
2N 2N
(6)
⎧
⎨ 1 for u = 0
α(u) = N (7)
⎩ 2 for u = 0
N
⎧
⎨ 1 for v = 0
α(v) = N (8)
⎩ 2 for v = 0
N

7
7
π (2x + 1)u π (2y + 1)v
tk (u, v) = α(u)α(v) f k (x, y) cos cos (9)
x=0 y=0
16 16
for u = 0, 1, 2 . . . 7 and v = 0, 1, 2 . . . 7, the inverse transform IDCT of block

tk (u, v) can be defined as

7
7
π (2x + 1)u π (2y + 1)v
f k (x, y) = α(u)α(v)tk (u, v) cos cos (10)
u=0 v=0
16 16
for x = 0, 1, 2 . . . 7 and y = 0, 1, 2 . . . 7. Here, k defines the block number that has

been processed, f k (x, y) is the kth block of input image of size 8*8 and tk (u, v) is
the kth transform block of block f k (x, y) [4, 5].
2.2.2 Discrete Sine Transform (DST)
Like DCT, DST is also a member of Discrete Trigonometry Transform (DTT), hence,
it also can be used two as transform input blocks. Like DCT, it has same forward and
inverse transform kernels as following:
r (x, y, u, v) = s(x, y, u, v)

π (2x + 1)(u + 1) π (2y + 1)(v + 1)
= α(u)α(v) sin sin (11)
2N 2N
Using Eq. (11), forward transform FDST can be defined as

7
7
π(2x + 1)(u + 1) π(2y + 1)(v + 1)
tk (u, v) = α(u)α(v) f k (x, y) sin sin
16 16
x=0 y=0
(12)
and its inverse transform IDST can be defined as

7
7
π(2x + 1)(u + 1) π(2y + 1)(v + 1)
f k (x, y) = α(u)α(v)tk (u, v) sin sin
16 16
u=0 v=0
(13)
Here, terms tk (u, v), f k (x, y), α(u), α(v) are similar to DCT [6].
2.2.3 Discrete Hartley Transform (DHT)
DHT is also one of the important transformations in image processing, it looks like
DFT but it does not generate complex numbers, hence, it is faster than DFT. For a
one-dimensional (1D) function f (x), DHT can be defined as [6, 7]
N −1

π (2x + 1)u
t(u) = f (x)cas (14)
x=0
N
and its inverse as

N −1

1 π (2x + 1)u
f (u) = t(u)cas , (15)
N u=0 N
where t(u) is the transform function, N is a positive integer, and casθ is defined as
casθ = sin θ + cos θ (16)
For 2D function, DHT has two variants of forward and inverse transform kernels
and can be defined as

π (2x + 1)u π (2y + 1)v
r (x, y, u, v) = cas + (17)
M N

1 π (2x + 1)u π (2y + 1)v
s(x, y, u, v) = cas + (18)
MN M N

π (2x + 1)u π (2y + 1)v
r (x, y, u, v) = cas *cas (19)
M N

1 π (2x + 1)u π (2y + 1)v
s(x, y, u, v) = cas ∗ cas (20)
MN M N
Equations (17) and (18) are known as normal forward and inverse transform
kernel, respectively, whereas Eqs. (19) and (20) are termed as forward and inverse
transforms for Separable 2D Discrete Hartley Transform (SDHT)n respectively.
Using Eq. (17) forward transform FDHT on image block, f k (x, y) can be applied as

7
7
π (2x + 1)u π (2y + 1)v
tk (u, v) = f k (x, y)cas + (21)
x=0 y=0
8 8
and its inverse transform IDHT can be defined as

1
7 7
π (2x + 1)u π (2y + 1)v
f k (x, y) = tk (u, v)cas + . (22)
64 u=0 v=0 8 8
Similarly, forward SDHT (FSDHT) can be defined using forward kernel as described
by Eq. (19)

7
7
π (2x + 1)u π (2y + 1)v
tk (u, v) = f k (x, y)cas *cas (23)
x=0 y=0
M N
and its inverse transform ISDHT can be defined as [8]

1
7 7
π (2x + 1)u π (2y + 1)v
f k (x, y) = tk (u, v)cas *cas (24)
64 u=0 v=0 M N
All the blocks generated by these transform are acting as input to quantization
process, where quantize blocks are formed for further processing.
2.3 Quantization
Quantization is a process of representing a large possible input set of value with

a smaller set. Quantization is an irreversible and lossy process because there is no
method by which input symbol can be reconstructed using quantized value.
In this phase, each transformed block is quantized using the following formula:

t(u, v)
QT B(u, v) = R OU N D , (25)
QT (u, v)
where QTB is termed as Quantized Transform Block, u = 0, 1, 2 . . . 7, v =

0, 1, 2 . . . 7. t(u, v) is the transform block obtain from Eqs. (9), (12), (21) or from
(23) and QT represents quantized matrix which can be either of following quantize
tables as shown below.
Because quantization is a lossy process, hence block t(u, v) cannot reconstruct in

decompression process but its equivalent block t (u, v) can be obtained by following
a formula:
t (u, v) = QT B(u, v) × QT (u, v) (26)

Table 1 3Quantize Table 1 [9]

16 11 10 16 24 40 51 61
12 12 14 19 26 58 60 55
14 13 16 24 40 57 69 56
14 17 22 29 51 87 80 62
18 22 37 56 68 109 103 77
24 35 55 64 81 104 113 92
49 64 78 87 103 121 120 101
72 92 95 98 112 100 103 99
Table 2 3Quantize Table 2 [10]

17 18 24 47 99 99 99 99
18 21 26 66 99 99 99 99
24 26 56 99 99 99 99 99
47 66 99 99 99 99 99 99
99 99 99 99 99 99 99 99
99 99 99 99 99 99 99 99
99 99 99 99 99 99 99 99
99 99 99 99 99 99 99 99
2.4 n-Modulus Method
The n-Modulus Method (shortly n-MM) transforms each input value into a number
that is divisible by n, this transformation will not affect the Human Visual System
(HVS).
n-MM reads each QTB block row by row and divides each input value by n and
adds or subtracts the reminder from the original pixel to get repeated pixel values.
The basic idea in n-MM is to check the whole pixel metrics and transform each pixel
into a number divisible by n. The reason behind using this method before encoding
is that an entropy encoding technique gives a high compression as the input values
have repeated data sequentially [11, 12].
Mathematically, forward n-MM is given as
QT B(u, v)
n−M M(u, v) = R OU N D (27)
n
and its inverse work as

QT B (u, v) = n−M M(u, v) × n (28)
2.5 Entropy Encoding
Encoding techniques are the key to data compression techniques. However, a base-
line or base value is required to identify how much amount of compression can be
achieved, this base value is given by entropy. Entropy is a probability-based method to
identify the average amount of information required representing each input symbol
and its mathematical statement can be defined as

n

H =− p a j logp a j (29)
j=1
Here, H is the entropy of input symbols a1 , a2 , a3 , . . . . . . . . . . . . , an , p(a j ) is the

probability of jth symbol in the sequence of input symbols. In the present paper for
final compression, RLE is used as an entropy encoding [13].
2.5.1 Run Length Encoding (RLE)
RLE is one of the simplest encoding techniques for compression; it is mainly used
for compressing sequential data. In most of the digital image, gray values of pixels
are repeated sequentially. Hence, RLE is suitable for such type of images. For every
repeated gray value, it only stores the gray value with their frequency rather than
storing that gray value every time [13].
2.6 Fidelity Analysis
The image compression model can be used for lossless compression as well as for
lossy compression; if it is used for lossy, then there is need to identify some criteria,
by which amount of information can be decided. There are two types of criteria for
such assessment [10].
2.6.1 Objective Fidelity Criteria
The criteria are given by mathematical functions that determine the amount of error
or noise arising in the image due to information loss. The terms to define these criteria
are MSE, PSNR, CR, CC, and NCC. For an input image f (x, y) and output image
f (x, y) with resolution M × N , these criteria can be defined as
e(x, y) = fˆ(x, y) − f (x, y) (30)


M−1 N −1
TE = fˆ(x, y) − f (x, y) (31)
x=0 y=0

M−1 N −1 2
1
MSE = fˆ(x, y) − f (x, y) (32)
M ∗ N x=0 y=0

N −1 2
1
M−1
er ms = fˆ(x, y) − f (x, y) (33)
M ∗ N x=0 y=0
M−1 N −1 2
x=0 y=0 fˆ(x, y)
S N Rms =
M−1 N −1 2 (34)
x=0 y=0 fˆ(x, y) − f (x, y)
PSNR = 20∗ log10 (MAX) − 10*log10 (MSE) (35)
Memory Required to store Original Image

CR = (36)
Memory Required to store Encoded data
M−1 N −1
x=0 y=0 ( f (x, y) − i mean )( f (x, y) − i mean )
CC = N −1 2
M−1 2 M−1 N −1
x=0 y=0 ( f (x, y) − i mean ) x=0 y=0 f (x, y) − i mean
(37)
M−1 N −1
x=0 y=0 ( f (x, y) × f (x, y))
NC = N −1 (38)
M−1
y=0 ( f (x, y))
2
x=0

Here, f (x, y) and f (x, y) are the pixel values at point (x, y) for original image

and decompressed image, respectively, i mean and i mean are the mean values for orig-
inal image and decompressed image, respectively, e(x, y) represents the error at
(x, y). TE represents the Total Error, MSE represents the Mean Square Error, er ms
represents Root Mean Square Error, S N R ms represents the Mean Square Signal to
Noise Ratio, PSNR represents the Peak Signal to Noise Ratio, and MAX represents
the maximum gray value of a pixel. For example, for an 8-bit image, MAX would
be 255 [10].
2.6.2 Subjective Fidelity Criteria
Subjective fidelity criteria are purely based on the visual properties of output image
f (x, y) and can be rated as follows based on the recognition of HVS [10] (Table 3).
Table 3 Subjective fidelity criteria [10]

Value Rating Description
1 Excellent An extremely high-quality image
2 Fine A high-quality image, but information losses not noticeable
3 Passable Quality of image is acceptable but degradation in image noticeable
4 Marginal A poor quality image and losses can be seen easily
5 Inferior A very poor image, but can be seen
6 Unusable Due to extreme degradation-level image cannot be seen
3 Results
The results vary on the basis of the input image and the value of n (from 1 to 10) in
n-MM algorithm. For each input image, results are categorized into two sections, one
is objective fidelity and other is subjective fidelity. Objective fidelity gives variation
in the objective parameter of input image, which are MSE, PSNR, Size, CR, CC, and
NC. Subjective fidelity shows the variation in vision quality on the basis of HVS. To
complete this research, the following platforms are used.
Hardware: Intel(R) Core(TM) i5-6200U CPU @ 2.30 Ghz with 8 GB RAM.
Programming Language: JAVA programming language, MATLAB.
Tools: NetBeans IDE 8.0.2 for Java, MATLAB R2010a for MATLAB, Microsoft
Office 2013.
Hardware: Intel(R) Core(TM) i5-6200U CPU @ 2.30 Ghz with 8 GB RAM (Table
4).
3.1 Object Fidelity Results for Lena.bmp
3.1.1 Variation in MSE for Lena.bmp
See Fig. 5.
Table 4 Sample input images

S. No Image name Image type Resolution Image size (KB)
1 Lena.bmp BMP 512*512 768
2 Pappers_grey.bmp BMP 512*512 768
3 Baboon.bmp BMP 512*512 768
n vs MSE
180
160
140
120
100
MSE
80
60
40
20
0
0 2 4 6 8 10 12
n
MSEDCTQ1 MSEDSTQ1 MSEDHTQ1
MSEDCTQ2 MSEDSTQ2 MSEDHTQ2
Fig. 5 Variation in MSE for Lena.bmp based on different transformation and quantization
n vs PSNR
50
40
30
PSNR
20
10
0
0 2 4 6 8 10 12
n
PSNRDCTQ1 PSNRDSTQ1 PSNRDHTQ1
PSNRDCTQ2 PSNRDSTQ2 PSNRDHTQ2
Fig. 6 Variation in PSNR for Lena.bmp based on different transformation and quantization
3.2 Variation in PSNR for Lena.bmp
See Figs. 6.
3.3 Variation in CR for Lena.bmp
See Fig. 7.
3.3.1 Variation in CC for Lena.bmp
See Fig. 8.
Fig. 7 Variation in CR for Lena.bmp based on different transformation and quantization
n vs CC
1.2
1
0.8
CC
0.6
0.4
0.2
0
0 2 4 6 8 10 12
n
CCDCTQ1 CCDSTQ1
CCDHTQ1 CCDCTQ2
CCDSTQ2 CCDHTQ2
Fig. 8 Variation in CC for Lena.bmp based on different transformation and quantization
3.3.2 Variation in NCC for Lena.bmp
See Fig. 9.
Similar results for Pappers_grey.bmp and Baboon.bmp are obtained.
3.4 Subjective Fidelity Results for Lena.bmp
See Table 5.
Here, SFV refers to Subjective Fidelity Value as described in Table 3.
n vs NC
0.08
0.06
NC
0.04
0.02
0
0 2 4 6 8 10 12
n
NCDCTQ1 NCDSTQ1
NCDHTQ1 NCDCTQ2
NCDSTQ2 NCDHTQ2
Fig. 9 Variation in NCC for Lena.bmp based on different transformation and quantization
Table 5 Variation in value of subjective fidelity, for Lena.bmp based on different transformation
and quantization
DCT + Q1 DST + Q1 DHT + Q1 DCT + Q2 DST + Q2 DHT + Q2
n 1 1 1 1 1 1
SFV 1 2 1 1 2 1
n 2 2 2 2 2 2
SFV 1 2 1 1 2 1
n 3 3 3 3 3 3
SFV 2 3 2 2 3 2
n 4 4 4 4 4 4
SFV 2 3 2 2 3 2
n 5 5 5 5 5 5
SFV 2 4 3 2 4 2
n 6 6 6 6 6 6
SFV 3 4 3 3 4 3
n 7 7 7 7 7 7
SFV 3 4 3 3 4 3
n 8 8 8 8 8 8
SFV 3 4 3 3 4 3
n 9 9 9 9 9 9
SFV 3 4 3 3 4 4
n 10 10 10 10 10 10
SFV 4 5 4 4 5 5
4 Conclusion and Future Scope
4.1 Conclusion
The present paper gives an analytical view of implementation and describes how the
size of input image is effected at the cost of quality of input image. The results shown
in this paper are categorized into two categories, objective fidelity and subjective
fidelity. Based on results obtained for objective, following conclusion can be carried
out.
• For all the input images taken in this research, DHT gives the best results with
respect to quality as shown in the table, but when compression ratio has priority,
then DCT is best among these three transformations as shown in the table. DCT
and DHT both are better than DST in all aspect. Hence, where quality has higher
priority, then order of good result is DHT > DCT > DST and for high compression,
the order will be DCT > DST > DHT. For the higher value of n order becomes DCT
> DHT > DST because DHT gives high CR than DST. However, if good quality
of image and good compression both are required, then DCT is more suitable. It
gives high compression at average degradation in input image (Tables 6 and 7).
• Mathematical transformation does not compress the input image without quanti-
zation and entropy encoding. In this research, two quantization tables Q1 and Q2
are used; results show that matrix Q2 nnot only gives high compression than Q1
but it also degrades input image in higher amount than Q1. Hence, if quality is
considered, then Q1 is more suitable than Q2, otherwise Q2 can be used for high
compression, as shown in Tables 8 and 9.
• The n-MM algorithm, used in this research, gives a range of degraded image from
low degradation to high degradation. Level of degradation depends upon the value
of n. As the value of n increases, the degradation level also increases and high
compression is also obtained for higher value of n, as given inTables 10 and 11.
n ∝ Compr ession ∝ Degradation
• The results of subjective fidelity are purely based on human vision and results
clearly show that input image is always visible and understandable even for higher
value of n, i.e., for n > 5.
• The results also show that the compression and degradation in input image not
only depends on algorithms used in the compression process but also depends on
input image (Table 12).
• Based on the value obtained for CC, it is clear that as the value of n increases,
deviation from original image also increases, but image remains visible for all
input test cases (Table 13).
242
Table 6 Comparison of Mathematical transformation DCT, DST and DHT for all the sample input images with respect to average MSE value
Input image MSE MSE MSE MSE MSE MSE
DCT + Q1 DCT + Q2 DST + Q1 DST + Q2 DHT + Q1 DHT + Q2
Lena.bmp Range 9.7–112.3 15.9–120 47.2–157.1 66.2–160.1 3.0–31.0 5–35.9
Average 59.5 66.9 124.5 131.5 19.8 25
Baboon.bmp Range 35.3–116.3 50.7–122.8 58.5–158.3 74.2–159 3.9–63.8 6.6–70.3
Average 84.1 91.2 125.8 135.8 43.5 52.19
Pappers_grey.bmp Range 14.6–113.4 19.3–120.0 47.5–157.0 65.5–159.9 3.8–34.2 6.0–38.9
Average 62.7 69.9 122.2 129.2 22.9 28.3
N. Agarwal and R. Purohit
Table 7 Comparison of Mathematical transformation DCT, DST and DHT for all the sample input images with respect to average CR value
Input image CR CR CR CR CR CR
Lena.bmp Range 7.9–37.9 13.3–43.7 2.96–11.9 3.6–19.7 1.9–13.6 2.89–22.4
Average 26.8 33.64 7.86 14.3 8.21 13.78
23 A Comparative Study of Various Approaches …
Baboon.bmp Range 3.5–36.0 6.5–45.2 2.4–11.54 3.18–18.9 1.3–7.15 1.5–14.1

Average 20.16 30.99 7.32 13.58 4.15 7.8
Average 27.8 34.4 8.2 15.1 8.02 14.07
243
244
Table 8 Comparison of quantization tables Q1 and Q2 for all the sample input images with respect to average MSE value
Input image MSE MSE MSE MSE MSE MSE
Lena.bmp Range 9.7–112.3 15.9–120 47.2–157.1 66.2–160.1 3.0–31.0 5–35.9
Average 59.5 66.9 124.5 131.5 19.8 25
Baboon.bmp Range 35.3–116.3 50.7–122.8 58.5–158.3 74.2–159 3.9–63.8 6.6–70.3
Average 84.1 91.2 125.8 135.8 43.5 52.19
Average 62.7 69.9 122.2 129.2 22.9 28.3
Table 9 Comparison of quantization tables Q1 and Q2 for all the sample input images with respect to average CR value
Input image CR CR CR CR CR CR
Lena.bmp Range 7.9–37.9 13.3–43.7 2.96–11.9 3.6–19.7 1.9–13.6 2.89–22.4
Average 26.8 33.64 7.86 14.3 8.21 13.78
Baboon.bmp Range 3.5–36.0 6.5–45.2 2.4–11.54 3.18–18.9 1.3–7.15 1.5–14.1

Average 20.16 30.99 7.32 13.58 4.15 7.8
Average 27.8 34.4 8.2 15.1 8.02 14.07
245
246
Table 10 MSE range comparison for all the sample input image
Input image MSEDCTQ1 MSEDCTQ2 MSEDSTQ1 MSEDSTQ2 MSEDHTQ1 MSEDHTQ2
Lena.bmp Range 9.7–112.3 15.9–120 47.2–157-10 66.2–160.1 3.0–31.0 5–35.9
Baboon.bmp Range 35.3–116.3 50.7–122.8 58.5–158.3 74.2–159 3.9–63.8 6.6–70.3
Table 11 CR range comparison for all the sample input image
Input image CRDCTQ1 CRDCTQ2 CRDSTQ1 CRDSTQ2 CRDHTQ1 CRDHTQ2
Lena.bmp Range 7.9–37.9 13.3–43.7 2.96–11.9 3.6–19.7 1.9–13.6 2.89–22.4
Baboon.bmp Range 3.5–36.0 6.5–45.2 2.4–11.54 3.18–18.9 1.3–7.15 1.5–14.1

247
Table 12 Input image

Input image MSE CR
comparison with respect to
Average Average
average MSE and average CR
value Lena.bmp 71.2 17.4
Baboon.bmp 88.8 14
Pappers_grey.bmp 72.6 17.9
Table 13 Input image

Input image CC
comparison with respect to
Average
average CC value
Lena.bmp 0.899
Baboon.bmp 0.826
Pappers_grey.bmp 0.903
4.2 Future Scope
In today’s era as the development and use of multimedia devices like smartphones,
LED TVs are increasing, the quality of images and videos are also increasing. In
current scenario, images and videos have billions of color as in full HD, 4 and 16 K
resolutions. As the resolution and color depth of images and videos increases, the
size of images and videos also increases. Hence, compression is required for storage
purpose as well as for transfer in the network. In future, compression can be used in
image compression as well it can be embedded into video compression like for online
streaming, where compression level can be selected as per requirement. The present
work can be extended for color images with some high computation complexity
because present compression technique will be applied on each color component
of color image. Compression and decompression are two separate phases in image
compression, hence, the present work can be used for cryptography in future, where
compression phase will be used for generating cypher image and decompression
phase will be used for getting original image from cypher image.
References
1. Gonzalez RC, Woods RE (2002) Digital image processing, 2nd edn. Prentice Hall
2. Lata A, Singh P (2013) Review of image compression techniques. Int J Emerg Technol Adv
Eng (IJETAE) 3(7):461–464
3. Kainmal AB, Manimurugan S, Devadas CSC (2013) Image compression technique: a survey.
Int J Eng Invent 2:26–28
4. Jassim FA (2013) Image compression by embedding five modulus method into JPEG. Signal
Image Process Int J (SIPIJ) 4(2):31–39
5. Chincholli P, Krishankumar E (2016) A comparative performance analysis of DCT, DWT-SVD,
DCT-DWT-SVD digital water marking algorithms. Int J Technol Res Eng 3(8)
6. Malini S, Moni RS (2014) Use of discrete sine transform for a novel image denoising technique.
Int J Image Process (IJIP) 8(4):204–213
7. Hu NC, Ersoy OK Generalized discrete Hartley transforms, China. In: 1991 international
conference on circuits and systems, June 1991. IEEE, pp 130–132
8. Watson AB, Poirson A (1985) A separable two-dimensional discrete Hartley transform. Nasa
Technical Memorandum 88203, December 1985
9. Kornblum JD (2008) Using JPEG quantization tables to identify imagery processed by software.
Digit Investig (Elsevier) 5:S21–S25
10. Li ZN, Drew MS, Fundamental of multimedia. Pearson Educational International. ISBN 0–
13–127256-X
11. Jassim FA (Member, IEEE) (2013) A novel steganography algorithm to hide a grayscale BMP
image in two grayscale BMP image for dual secrecy. In: 2013 2nd national conference on
information assurance, pp 73–77. IEEE
12. Jassim FA, Quassim HE (2012) Five modulus method for image compression. Signal Image
Process Int J (SIPIJ) 3(5)
13. Salomon D (2007) Data compression: the complete reference, 4th edn. Springer, Berlin
Chapter 24
Practical Design Considerations
of DC/DC Converter Used in MPPT
for Solar PV Systems
Kusum Lata Agarwal and Dhanraj
1 Introduction
The solar PV system not only consists of PV panels but also uses a power elec-
tronic converter for connecting its output to the grid or to the local load. The power
electronics converters normally used are, the DC–DC converter and the DC–AC
inverter for AC conversion. To extract the maximum power from the SPV module,
the Maximum Power Point Tracking (MPPT) algorithm is incorporated with the
DC–DC converter to step up/down the level of the solar PV array output voltage.
The power semiconductor devices are used as switches where the internal resistance
switch and internal resistance of practical inductors introduce new challenging prob-
lems like power loss, electromagnetic interference (EMI). Due to these problems,
efficiency and power quality go down. To mitigate these issues, it is essential to create
efficient DC–DC converter and effective algorithms for the MPPT.
The boost converter is used to uplift the level of DC voltage as per requirement. In
literature, many authors have been worked out on simulation and hardware realization
of the boost converter with and without solar photovoltaic application.
Kollimalla et al. [1] have presented the hardware prototype. They found the
dynamic performance and stability of DC–DC boost converter using variable pertur-
bation size adaptive P&O MPPT algorithm for rapid changes in solar insolation. In
this algorithm, they used current perturbation, adaptive control and variable pertur-
bation. The converter efficiency was not available in terms of numerical number but
proposed in terms of dynamic performance of DC–DC boost converter with rapid
K. L. Agarwal (B) · Dhanraj

Department of Electrical Engineering JIET, Jodhpur, Rajasthan, India
e-mail: kusum.agarwal@jietjodhpur.ac.in
Dhanraj
e-mail: dhanrajchouhan013@gmail.com

https://doi.org/10.1007/978-981-15-8820-4_24
252 K. L. Agarwal and Dhanraj
changes in irradiation. Singh et al. [2] have been demonstrated the boost converter-
based solar PV fed brushless DC motor driven the water pump. In this, they have
used the DC–DC boost converter to achieve the MPPT of the PV array. The input PV
parameters (Panel voltage and panel current) have sensed and fed to the INC MPPT
algorithm to generate the desired PWM signal in MATLAB/Simulink. In this, the
MATLAB-based simulation work has been carried out. They did not discuss the
efficiency of the DC–DC boost converter.
To operate the AC appliances using Solar PV array, the voltage of DC link √capac-
itor at the input side of the inverter should be greater than or equals to 2 2 time
of phase voltage, as determined by Popa et al. [3]. In this, they have investigated
that DC–DC boost converter can be the best choice to reduce the number of SPV
modules are being connected in series. Balamurugan et al. [4] have proposed the
design of boost converter for PV application. In this, to track the MPP point of solar
PV panel, the Perturb and Observe (P&O) method has been used. The ATmegha328
microcontroller has used to extract the maximum power from the solar PV panel.
Saxena et al. [5] have proposed an integration of solar photovoltaic with battery to
single-phase grid in which the boost converter has used to step-up the DC voltage for
the dc-link capacitor. Sathya et al. [6] have designed and implemented a 12 V/24 V
boost converter for the solar-powered LED lighting system. The output voltage has
controlled by the use of a closed-loop feedback system. In this, they have observed
a 95% efficient boost converter in hardware realization.
Many authors have tried to address the DC–DC converters interfacing a load
with either solar PV panel/module/array or rectified DC supply in order to extract
the maximum power from the given source. But is found that no single literature is
there to present the hardware realization of the proposed DC–DC boost converters.
The objective of this paper is to propose step-by-step design method to develop the
hardware prototype for the proposed DC–DC boost converter.
2 Basics of Boost Converter
As the name implies, a step-up (Boost) converter produces a higher average output
voltage than the DC input voltage V PV . Many times, this converter can be used
to match the load impedance with impedance of solar photovoltaic panel (Ropt =
V mp /I mp ) to transfer the maximum power or to operate the SPV panel at MPP point.
Figure 1 shows the SPV panel and this is interfaced to RL load with the help of DC–DC
boost converter. The boost converter consists of an inductor at input, semiconductor
switch (it could be BJT, IGBT, and MOSFET), diode and a capacitor.
24 Practical Design Considerations of DC/DC Converter … 253
Ipvav L
Ipv
VL Io
+
PV Panel
CT S1 Cout RL Vo
Vpv G
-
Fig. 1 Working of boost converter during DT s time
IPVav IPV L D1
VL Io
+
PV Panel
CT Cout RL Vo
VPV
-
Fig. 2 Current paths during (1 – D)T s time in boost converter
2.1 Working of Boost Converter
During ON-Time
• During ON-time, the terminal current I PV flows from PV module to an inductor
and through the switch-Q as depicted in Fig. 1.
The capacitor Cout continues to discharge through the load while the inductor L
charges by extracting the energy from the SPV module.
During ON-time, the voltage across the inductor is V PV and current through it,
rises linearly from a non-zero value to a peak value.
During OFF-Time
• During off-time, the inductor current flows continue and therefore the polarity of
inductor voltage reverses and diode D conducts. And this current will charge the
capacitor C out and supply the load current I o .
• The voltage across the inductor is V PV – V o . As V o > V PV , the voltage across the
inductor is negative and this is expected because we need the positive portion to
balanced out by negative portion.
• The current through inductor will decay because this is discharging now into the
capacitor and discharging into the load. The capacitor is charging up.
• The average inductor current is the terminal current I PV (Fig. 2).
Input–Output Voltage Relationship:

V pv × DTs + V pv − Vo × (1 − D)Ts = 0

1
V o = V PV
1− D
Input–Output Current Relationship:

The average current through the capacitor should have to be zero. Meaning that there
is no charge building up in capacitor or capacitor is in steady state. Discharging
capacitor current is taken as negative value.
(−Io )DTs + (I P V − Io ) × (1 − D)Ts
Io = IPV (1 − D)
On dividing the equations,

1
Vo V P V 1−D VP V 1 RT
RL = = = =
Io I P V (1 − D) I P V (1 − D)2
(1 − D)2
RT
RL =
(1 − D)2
2.2 Effect of Boost Converter on MPPT
The terminal resistance
R T = R L (1 − D)2
The terminal resistance or input resistance to the boost converter can be changed
by varying the duty ratio. Let us draw the I–V characteristic and power curve of PV
module as depicted in Fig. 3.
Ipv
Io
VPV Boost
Vo RL
RT Converter
Duty
Fig. 3 PV module with MPP block for boost converter
Fig. 4 PV module I–V 1

characteristics for boost IPV RL @ D =1 1
converter Ropt 1
@D=
R
VPV
If duty D = 1, then.
RT = 0, i.e., vertically short circuited and the load line is along with y axis.
If duty D = 0, then.
RT = RL .
With Boost converter the following point on PV curve can be traced. There is
limitation with load line; consequently, there is limited load line (Fig. 4).
Hence, in boost converter the reachable points are lying between the load line
corresponding to D = 0 and corresponding to D = 1 as shown in Fig. 5. From the
above, it can be concluded that the maximum power point is not traced with Boost
converter.
Fig. 5 PV module I–V 1

characteristics with limited 1 @ D=0
@ D=1 RL 1
load line for boost converter Ipv RL RT
@ MPP
Vpv
3 Design of Boost DC–DC Converter
The estimation of inductor L and capacitor C, where f s is the switching frequency

of boost converter; I L is the average inductor current; I L is ripple content in the
inductor current; I L Max is a maximum value of inductor current; I L Min is a minimum
value of inductor current; V o is ripple content in the capacitor voltage.
Step 1: Consider the values of Pmpp , V mpp , I mpp , I sc .

Step 2: If converter has zero internal losses then output voltage will be
Vo2
Vin Iin = Vo Io = = Pmpp
RL
Vo2
= Pmpp ⇒ Vo = Pmpp ∗ R L
RL
Step 3: For given output voltage, the duty of converter will be
Vin Vin
Vo = ⇒ D =1−
1− D Vo
Step 4: The inductor current is given by
Vin
IL =
(1 − D)2 R L
If the change in inductor current IL is 5% then
I L = 0.05 ∗ I L
Since,
Vin D
I L =
L fs
Determine L from the above expression.
Step 5: Maximum and Minimum Inductor Currents.

The maximum current through the inductor in Boost DC–DC converter is given
by
I L
I L max = I L +
2
The minimum current through the inductor in Boost DC–DC converter is given
by
I L
I L min = I L −
2
Step 6: Value of Output capacitance, if the ripple is not more than 1% then
D
C=
R L f s V
Vo
o
3.1 Designing of Practical Inductors (L) and Selection

of Capacitors (C) for Proposed Converter Topologies
The practical inductor consists of a wound conductor coil on a ferromagnetic material.

This combination yields an inductance (L) that offers reluctance to a change in
current, and therefore the current through an inductor cannot change instantaneously.
The rate of change of current through an inductor (diL / dt) is determined by the
inductance and the voltage dropped across the inductor, given by the expression: V
= L * diL / dt. Furthermore, the use of ferromagnetic material as the inductor core
allows energy to be stored in the inductor. When a positive voltage has appeared
across the inductor, the current increases and meanwhile energy is dumping into the
inductor. It is an essential characteristic that makes the inductor useful in the DC–DC
converter since it acts as both a current-ripple filter and an energy-storage element.
When the switch is closed, the current flowing to the load increases and energy is
also stored in the inductor. When the switch is opened and the output is disconnected
from the input, the stable output current is maintained by drawing energy from the
inductor. Since inductance determines the diL / dt, its value is selected to achieve
desired limits to the ripple current (I ripple ), providing a steady output current. The
inductor can only hold a finite amount of energy before the ferromagnetic material
will saturate, the inductance decreases, and ripple current increases. When making an
inductor selection, it is important to check that the current at which the core saturates
(I sat ) is greater than the application’s peak inductor current, (I pk = I out + I ripple /2).
In this thesis, all topologies work in continuous conduction mode (CCM). For CCM,
the practical value of inductor should be greater than the minimum value of required
inductor values.
3.1.1 Design of Inductors
The inductor is the energy storing element and practical inductor has some value of
its wire resistance. The wire resistance should be very small as possible therefore;
the proper design of practical inductor plays an important role in MPP of SPV panels
or modules. The following steps are used to design the inductors used in DC–DC
converter.
Step 1 (Inductor Value): the value of the inductor has to be calculated. The
calculation of inductors is done for DC–DC converter topologies.
Step 2 (Area Product): From the current waveform through the inductor, the energy-
storage requirement for the inductor is estimated. The energy that needs to be stored
in the inductor is given as
1 2
EL = LI
2 m
where E L is the maximum energy in joules that needs to be stored in the inductor
and I m is the peak inductor current in amperes. Based on the energy to be handled
by the inductor core, the area product is given as
2E L
A p = Ac Aw =
K w K c J Bm
where:
K w = 0.6 for single-winding inductor and 0.3−0.4 for multiple-winding inductors
Kc = 1
J = Current density for copper = 3 × 106 A/m2

Bm = Flux density in Tesla = 0.25 T for ferrites, 1 T for CRNGO and 1.2 T for
CRGO.
The selected core should have an area product greater than that calculated by the
above equation.
Step 3 (Permeance): The permeance of a core with air gap lg is given by
μo μr Ac
∧=
lm + μr l g
where:
Ac = core cross-sectional area of the selected core (in m2 )
µr = relative permeability of the selected core material, for ferrite core it is generally 2000
l g = introduced air gap length (in meter)
lm = magnetic path length of the selected core (in meter)
Step 4 (Number of turns): The number of turns required for desired value of
inductance is given by

L
N=
∧
Step 5 (Gauge of Wire): the cross-section area of the wire can be calculated as
Irms
a=
J
where, I rms is the RMS value of the current flowing through the inductor which is
equal to I m /K c and K c = I m /I o . The gauge of wire (SWG) selected should have a
cross-section area that is greater than that calculated by the above expression.
Step 6 (Available Window Area check): The inequality AW K W > aN should be
satisfied or else repeat the calculation for the number of turn and gauge of wire after
choosing the next bigger core. Note that the value ‘a’ used for checking the above
inequality should be the actual cross-section area of the wire and not the calculated
value in step-5.
3.2 Selection of MOSFET Switch and Driver Circuits
Most DC–DC converters are designed for continuous current operation. Note that
as the switching frequency increases, the minimum size of the inductor to produce
continuous current and the minimum size of the capacitor to limit output ripple both
decrease. Therefore, high switching frequencies are desirable to reduce the size of
both the inductor and the capacitor. The tradeoff for high switching frequencies is
increased power loss in the switches. Increased power loss in the switches means
that heat is produced. This decreases the converter’s efficiency and may require a
large heat sink, offsetting the reduction in the size of the inductor and capacitor. The
selection of power MOSFET based on that the rdON should be very low and voltage,
current ratings of MOSFET.
Electrical isolation is provided with the use of integrated circuit (IC) package
TLP250, TLP350 manufactured by TOSHIBA. This IC consists of a GaAlAs light
emitting diode and an integrated photo detector. The schematic and pin configuration
diagrams of TLP250 are depicted in Fig. 6.
Figure 6a clearly shows the input LED side and the receiving photodetector as well
as the totem-pole driver stage. Pins 1 and 4 are not internally connected to anything
and hence are labeled NC meaning no connection. Pin 8 is V CC —the positive supply.
Pin 5 is GND—the ground supply or the return path for the driving power supply.
The supply voltage must be at least 10 V. The maximum voltage is dependent on
the operating temperature. If the temperature is lower than 70 °C, up to 30 V can
be used. For temperatures between 70 and 85 °C, up to 20 V can be used. However,
there should not be a need to use higher than 20 V anyways (Fig. 7).
The Pins 2 and 3 are the inputs to the LED, anode and cathode, respectively.
Like regular LEDs, it has an input forward voltage and a peak forward current. The
(a) IC (b)
1 8
VCC
8
IF (Tr1) 2 7
2+ VO
7
VO 3 6
3- 6
(Tr2)
5 4 5
GN
Fig. 6 a Schematic diagram and b pin configuration of TLP250
+15V
330Ω 1 8
Rg
PWM Signal 2 7 +
from 3 6 0.1μF
4 5 -
Fig. 7 Gate driver circuit using TLP-250

forward voltage will typically be between 1.6 and 1.8 V. The forward current should
be less than 20 mA. The threshold input current for output transition from low to
high is typically 1.2 mA but may be as high as 5 mA. Thus, 10 mA current should
be good. Even though pins 6 and 7 are shown to be internally connected, the output
should be taken from pin 6 as the image—datasheet—shows pin 6 labeled as V o
(Output). The output voltage will tend to rise to supply voltage when high (it will
actually be slightly lower) and fall to ground level when low.
3.3 Design of Snubber Circuit for MOSFET
There is a danger of exceeding the voltage and the current rating of the MOSFET
switch during turn-OFF and turn-ON instants, respectively. If there is an inductive
load in the source side or if there is a significant amount of lead inductance associated
with the source or drain terminal, then when MOSFET is being turned OFF, the
current through the device could decrease rapidly to zero within the fall time of the
device. As a consequence, a large voltage spike due to Ldi/ dt will occur across the
device and cause V DS of the MOSFET to have a large spike during the fall time. This
may damage the device. Therefore, it becomes necessary to limit the voltage spike
during fall time.
On the other hand, during turn-ON of the device, due to the presence of any
capacitive load or parasitic capacitance across the switch, there will be a huge surge
current through the device which could damage the device. Therefore, it becomes
mandatory to limit the current spike through the device during turn-ON.
The Snubber circuits modify the device switching characteristics and in doing
so, reduce the device transients. The main task of the Snubber circuit is to absorb
energy from the reactive elements. The parasitic capacitance and inductance cause
large turn-on/off oscillation in the PWM input signal. The Snubber circuit improves
the input signal shape.
In the MPPT operated DC–DC converter, the input current is fairly constant and
it can be seen that the switch (MOSFET) is connected in the line or between the
line and ground terminal. When the MOSFET is turned OFF, the input current or
switch current iT will fall to zero in a period of time corresponding to the fall time t f
mentioned in datasheet of MOSFET. Due to this, the voltage across MOSFET V ds
will shoot up to a large value because of the Ldi/ dt phenomenon where L could be the
lead inductances associated with the drain and source leads. Therefore, it is required
to limit the voltage across the drain-source of the MOSFET and it should be made to
rise gradually during the fall time t f as depicted in Fig. 8b. To decrease the rate of rise
of voltage V ds , one can connect a capacitor C s across MOSFET as indicated in Fig. 8a.
Sometimes this Snubber circuit is also called as shunt Snubber because the capacitor
C s is connected in shunt with the device or MOSFET. Without loss of generality, one
can assume that the switch current is falling linearly as shown in Fig. 8c during the
t f time and the drain-source voltage across the device is rising linearly as depicted
in Fig. 8c during the fall time when the shunt Snubber is connected. From Fig. 8a,
id IPV
Ds Rs
t
Voff
vds
Cs
t
tf
(a) (b) (c)
Fig. 8 Turn-off or shunt Snubber circuit for power MOSFET
i P V = i d + i cap
where id is the current through the MOSFET and icap is the current through the shunt
capacitor. During the fall time period of t f , the current id through MOSFET is given
by
t
id = i P V − i P V
tf

t
id = i P V 1 −
tf
The current through the capacitor is given by

t t
i cap = i P V − iPV 1 − = iT ×
tf tf
The voltage V ds across the device is the same as the voltage across the capacitor
C s . therefore
1
vds = i cap · dt
Cs
1 t
vds = iPV × · dt
Cs tf
The voltage across MOSFET or capacitor can be obtained during fall time by
integrating the above equation within the fall time and then applying the boundary
condition. At the end of fall time, the voltage across the MOSFET should be VPV or
Vin then the value of capacitor will be
iPV t f
Cs =
2Vin
The problem of voltage spike during turn-off process of the MOSFET is resolved
by the circuit of Fig. 8a but it will create a serious problem during the turn-ON of the
MOSFET. When the MOSFET is turned ON again, the capacitor C s will discharge
through MOSFET and will result in a large current spike through the MOSFET
which can damage the power MOSFET. Therefore, to control the current through
MOSFET, a resistor Rs is introduced in series with capacitor C s.
The resistor Rs should provide the function of current limiting only during turn-
ON of the MOSFET. However, during turn-OFF of MOSFET, Rs is not needed.
Therefore, to deduce the dissipation in Rs during turn-OFF, a diode is placed across
Rs as shown in Fig. 8b so that Rs comes into effect only during turn-ON when C s
discharges through Rs and MOSFET.
3.4 Choose a Snubber Capacitance Value Which Meets Two

Requirements
(1) It can provide a final energy storage greater than the energy in the circuit
inductance
L i2
1
C V 2 > 21 L i i 2P V , and therefore Cs > Vi 2P V .
2 s off off
Where, L i is circuit inductance and V off is the voltage across the device when it
is turn-OFF.
(2) It produces a time constant with the Snubber resistor that is small compared
with the shortest expected on-time for the MOSFET switch.
tON
Rs C s <
5
4 Selection of Snubber Resistor Rs
When the MOSFET is turned on, it should carry the load current plus the capacitive
current discharge from C s which is equal to V off /R. therefore,
Voff
+ i P V < Idm
Rs
where, I dm is the maximum drain-current rating of the MOSFET.

Rearranging the above inequality, the following inequality is obtained
Voff
Rs >
Idm − i P V
It is also important to ensure that the capacitor discharges fully before the next
charging when the MOSFET turn-OFF. Therefore, there is a minimum duration of
time during which time the MOSFET should remain ON so that the capacitor can
fully discharge. Therefore
tON > 5Rs Cs
where, t ON is the minimum time for which MOSFET should remain in the ON-state.
Rearranging the above inequality, the following inequality is obtained:
tON
Rs <
5Cs
From the above two inequalities, the range of choice of Rs is given by
Voff tO N
< Rs <
Idm − i L 5Cs
The power rating of Snubber resistor is given by
1
PRs = 2
Cs Voff f sw
2
where, f sw is the switching frequency of the MOSFET.
5 Conclusions
The significant contributions of the presented paper are as follows:

1. The need of DC–DC converter in between solar PV module and resistive load
has been investigated through mathematical calculation.
2. The investigation on need of DC–DC converter revealed that DC–DC converter
with MPPT technique modifies the effective resistance being faced by Solar PV
module.
3. The parameters (Value of L, C and Duty) of the proposed DC–DC converters
have been determined keeping in mind that these converters must work in linear
region and in continuous conduction mode (CCM).
4. The most important parameter of DC–DC converter interfacing SPV module with
load is the internal resistance of inductor. The internal resistance of inductor may
restrict the maximum power transfer. Hence, the internal resistance of inductor
can be lowered by using Litz wire or large cross-sectional copper wire.
5. The internal resistance of high speed switch (MOSFET/IGBT) must be low. The
careful selection of power MOSFET plays a major role in energy extraction from
SPV module.
6. To drive the Power MOSFET, a robust optocoupler (TLP-250) based driver circuit
has been developed in the laboratory. The ringing effect has been minimized by
connecting a snubber circuit across the power MOSFET.
References
1. Kollimalla SK, Mishra MK (2014) Variable perturbation size adaptive P&O MPPT algorithm
for sudden changes in irradiance. IEEE Trans Sustain Energy 5(3):718–728
2. Singh B, Kumar R (2016) Solar PV array fed brushless DC motor driven water pump. In: IEEE
6th international conference on power systems (ICPS), March 2016
3. Popa D-L, Nicolae M-S, Nicolae P-M, Popescu M (2016) Design and simulation of a 10 MW
photovoltaic power plant using MATLAB and Simulink. In: IEEE international power electronics
and motion control conference (PEMC), September 2016, pp 378–383
4. Balamurugan M, Narendiran S, Sahoo SK, Shanmugam PK, Naidu RC (2016) Design of boost
converter for PV applications. In: Proceedings of fifth international conference on soft computing
for problem solving, April 2016, vol 2, pp 619–628
5. Saxena N, Singh B, Lal A, Vyas (2017) Single-phase solar PV system with battery and exchange
of power in grid-connected and standalone modes. IET Renew Power Gener 11(2):325–333
6. Sathya P, Natarajan R (2013) Design and implementation of 12V/24V closed loop boost converter
for solar powered LED lighting system. Int J Eng Technol (IJET) 5
Chapter 25
Machine Learning-Based Solar Energy
Forecasting: Implications on Grid
and Power Market
Chandershekhar Singh and Ajay Kushwaha
1 Introduction
In modern world, electrical power is a primary need and is counted in basic needs of
society. With the increase of residential and industrial loads, generation (or supply)
must increase at same rate to balance demand-supply. Conventional power plants
(thermal) have been employed since various decades to satisfy energy requirements.
It was a thinking that coal (fuel) is in abundance and it will never get exhausted.
Apart from it, supply from conventional generation increases greenhouse gas (GHG)
to atmosphere. However, there are methods to control/reduce GHG in conventional
plants but it will be better option that fossil fuel plants may be replaced by renewable
energy resources.
As of now, solar and wind energy generations are developed technology. They
have been installed and maintained on a large scale for power production. Despite of
intermittent nature of output power, lot of resource on their design and development
are considered by various nations.
Solar energy is vast and spread over every part of land and water on earth, but it
is not constant. It varies with respect to place and time due to geographical factors,
atmospheric conditions, etc. Global solar power installation has reached above 500
GW, with annual increase of 100 GW only in year 2018. India has also increased its
installation of 10.8 GW in year 2018, and adding total solar capacity to 33 GW [10].
Now a day’s most of electrical grids are centralized in various nations. They
transfer power from power plants (mainly thermal) to consumer end; however, decen-
tralized dispatch scenario is increasing significantly due to extension of renewable
generation (or distributed generation). There is a need to increase electricity genera-
tion amongst grid at different levels and this may be challenging to grid management.
C. Singh (B) · A. Kushwaha

Department of Electrical Engineering, JIET, Jodhpur, India
e-mail: chandershekhar333@gmail.com

https://doi.org/10.1007/978-981-15-8820-4_25
268 C. Singh and A. Kushwaha
Fig. 1 Low voltage ride

through requirement for grid
connection
Fig. 2 Line voltage and frequency operating limits of generating unit
Fig. 3 Voltage and power

factor operating limits for
generation unit
The challenge for electrical grid operators will be to optimize generation location
and dispatch to balance demand with energy supply and also maintain stability.
In solar technology, flat-panel PV, Concentrated Solar Power (CSP) and concen-
trated PV (CPV) system are mostly employed. They have ability to compete with
25 Machine Learning-Based Solar Energy Forecasting … 269
Table 1 Forecasting methods for the optimization of power output/Irradiance

Author Forecasting Methods INPUT Parameters OUTPUT
time horizon
1zgi et al. [1] 0–60-min Artificial Neural 750 Wp solar PV Solar power
ahead Network—Levenberg Panel—Ambient and
Marquardt Back- cell
Propagation temperature
Algorithm —Diffuse solar
Irradiation
Kardakos Day and Seasonal- Solar radiation PV power
et al.[2] Intra-day Autoregressive
ahead Integrated Moving
Average
—Artificial Neural
Network with
multiple input
Mellit et al. [3] Hourly Artificial Neural Solar irradiance Power of
Network—Multi —Air temperature PV
Layer Perceptron modules
Architecture
Yona et al. [4] 24-h ahead Based on insolation Weather reported PV power
forecasting by data (Humidity, cloud output
utilizing weather amount, and
reported data, fuzzy temperature)
theory, and Radial —Insolation forecast
Basis Neural data with fuzzy
Network
Almonacid 1-h ahead Dynamic Artificial Global solar PV power
et al. [5] Neural Network: irradiance output
Non-linear —Air temperature generator
Autoregressive —PV power output
Neural Network Module
(continued)
fossil fuel-based production in near future. However, natural variability (intermit-

tency) of the solar power and high cost of energy storage raises concerns regarding
reliability and feasibility of this technology.
2 Technical Issues with Grid
Grid interconnection (loading) and disconnection (unloading) of generation units is

one of the technical issues that needs to be considered for power dispatch.
The incoming generation unit and existing power-sharing (supplying) system have
to adjust their power output such that dynamics at the point of interconnection would
remain within limits. If new power generation unit is added it should synchronize
Table 1 (continued)
Author Forecasting Methods INPUT Parameters OUTPUT
time horizon
Cornaro et al. 24-h ahead Statistical Model Statistical model: Global
[6], (ST) and Model STNNS horizontal
Output Statistics —Ordinal day irradiance
(MOS) based on an number
ensemble Artificial —Daily irradiation
Neural Networks by —Average daily
master optimization temperature
process —Clearness index
MOS model:
ECMWF-MOSNNS
—Ordinal day
number
—Daily irradiation
—Average daily
temperature
—Clearness index
—Daily irradiation
prediction provided
by
Lima et al. [7] 24-h ahead NWP-WRF model by Weather forecast Surface
using Artificial outputs supplied by solar
Neural Networks WRF model irradiance
—Observational
ground data
Hossain et al. Day ahead Extreme Learning Daily average solar PV power
[8] —1-h ahead Machine (ELM) irradiance
—Ambient
temperature
—Module
temperature
—Wind speed
—PV output data
Mellit, A. Hourly and Adaptive feed-forward Solar Insolation time Hourly
MassiPava, V. Daily back-propagation series Insolation
Lughi [9] network (AFFNN)
its voltage, frequency, phase sequence as per grid codes. It also has to maintain his
loading gradually and reach up to sharing limit of power allocated. It is so because
ramp rate of loading has to maintain in limits, this ensures safety of generation
equipments. The active and reactive power controlled by DC to DC converters and
inverters at point of connection. These converters maintain parameter dynamics in
range. In wind energy system, back to back converters are used to control dynamics.
Generation unit is also required to operate in normal and faulty conditions. In
case of high transients/dynamics of grid or generation parameters, generation unit is
made to trip. There are standard conditions when generation could remain connected
to grid. This is fault ride through requirement.
It can be observed from the figure that generation unit may remain connected
if fault is cleared within 150 ms. Also if voltage due to fault restore its operating
magnitude in specified time (as straight line), then generation remains in operating
mode, otherwise unit may be disconnected from grid [11].
Frequency and voltage operating range must be in limits for continuous operation
of generation unit. In the figure, standard line voltages are 110, 220 and 380 volts and
their range of continuous operation is 100–123, 210–245 and 360–420, respectively,
if frequency is maintained 49–50 Hz. Deviation from standard magnitudes increases
dynamics/disturbance in system, hence more deviation in parameters will lead to
more disturbance of system. This gives less time for system to withstand disturbances.
Hence system becomes prone to tripping of generation unit [12].
These disturbances in voltage and frequency create flicker. When load changes
are more dynamically and demand–supply difference (changes) becomes fast then
power fluctuation arises. This creates power quality problems.
Uneven demand pattern and less regulated generated power (especially in case of
renewable) create grid instability in terms of frequency limit violation.
Reactive power is regulated to maintain system voltages within limits. In case of
faults, voltage is restored by increasing reactive power to the system. The reactive
power supply is provided by generator units, static capacitor banks, FACTS, etc.
In the figure, generation units supplying inductive VARs increases voltage
magnitude and capacitive VARs decreases voltage [12].
Now all problems start with demand–supply difference, it is needful that load
demand information must be specified. As the consumer is free to vary its load,
therefore, there is only left provision to study its demand pattern/behavior. Hence
power is regulated with respect to demand pattern of consumer.
It is a common practice to forecast demand behavior and manage supply resource
accordingly.
If power forecasted becomes same to that needed by consumer, then generator,
then load flow analysis can be carried out for stability analysis of Grid easily. Also
there will be better stability condition as power injection is defined and maintained
at a same level.
In thermal, energy/power output is controlled/regulated and fed to injection point
of grid in order to obtain stable grid interconnection.
If solar power could be predicted, then interconnection will be more reliable.
For controlling power, power electronics devices can be employed to obtain stable
interconnection.
For solar DC to DC converters, grid inverter controls parameter of solar power
with respect to Grid parameters. The power electronic devices selected must have
burden as that of installed solar capacity.
3 Forecasting and Power Market
Solar energy is highly dependent on weather conditions especially cloud structure

and day/night cycles. Clouds cause significant ramps in solar insolation and PV power
output. Therefore, integration of electricity produced by solar power systems requires
accurate energy evaluation and time horizon forecasts. Overall energy production
pattern does not follow the energy forecasted.
Solar forecasts on multiple time horizons play a fundamental role in
1. Storage management of PV systems.
2. Control systems in buildings, such as solar thermal plants
3. Power scheduling and regulation.
It allows grid operators to study load pattern in order to optimize power wheeling
and plan maintenance activities at the production sites and take necessary measures
to protect production from extreme events.
Depending on the purpose, different information are needed, such as
1. Long-term historical data of expected energy yield (to assess sites where solar
power is abundant)
2. Real-time data set of energy yield (optimizes real-time energy production
management)
3. Forecasted site irradiance data.
4. Real-time datasets on weather conditions.
If generation accuracy increases, it can be compared with conventional generation
resources that are scheduled and dispatched for power market.
Overall, accuracy in short-term forecasting yields:
(i) Exact Supplier (power) bidding in market and forecast returns from it.
(ii) Exact change in grid condition/grid parameter (such as frequency, reactive
power, voltages) for which a load flow study program can be run and take
an observation of grid parameter, which leads to grid stability.
Forecasting of supply is needed for ensuring power to be traded in power market.
The Generation supply bids have power (in MW), time span (for dispatching) and
price of power in bidding offer. Better forecasting ensures estimation of real-time
magnitude of power. More the reliable prediction, more reliability to market players.
Hence economic efficiency/incentives of transactions increase. There are various
methods for estimation/predication of power generated with respect to state variables
such as global irradiance, humidity, cloud, etc.
Apart from grid interconnection, grid stability is also essential; stability refers to
frequency regulation, voltage regulation, energy imbalance, contingency analysis in
case of outage, etc. Frequency regulation is maintained by demand supply balance.
Hence reserves are needed to balance them. These reserve services are also traded
apart from real power market.
Price bidding is normally for centralized market. In Indian system, discoms and
few retailers along with some big industries are “buyer” and all generations units
are “seller.” Here power bid is placed by every generator (government/private) in

market and buyers fulfill its power requirement and place his bid (prices and MW).
Market settlement is made by government entity and then transmission lines are
ensured/scheduled for bulk power transmission [13].
The generation price bid must be such that it becomes economical and reliable to
buyer. Here cost of power is overall components of various costs, such as capital cost
of installation, operation and maintenance, along with market returns (profit) whereas
reliability depends on technology of generation [14]. Solar technology produces
intermittent power that is not reliable in power market (in terms of fixed MW placed).
Hence solar power bid can be placed in market as minimum limit value of
range in which generated solar power varies. But as difference between minimum
and maximum is very large, therefore, this is another challenge to estimate about
magnitude of power placed in power market.
If forecasted solar power is close to real power produced then power placed in
market will be a curve of “power vs. time.” This variation of power can be managed
by system operator in its load flow studies. At last, payment for power transaction
will be on the basis of total energy transaction at the end of schedule.
4 Machine Learning-Based Solar Generation Models
As power output of solar PV modules depends on atmospheric and geographical

conditions, its power output can be predicted. There are various scenarios in which PV
modules behave in different way. These behaviors are modeled by different methods,
some of these methods are persistence method, statistical method, machine learning
approach and hybrid approach [15].
In persistence model, solar insolation and climatic conditions of each day are
considered to be same. Daily pattern of solar irradiance throughout the day of same
geographical location is fixed. It does not depend on change of climatic conditions
such as temperature, pressure, cloud variation, humidity, wind flow, suspension parti-
cles of air, etc. This method gives standard data that may be different from power
output of module of next coming day. These data are used as benchmark for the
analysis of power output when metrological data are considered in input.
In statistical model, more than one variable is considered for evaluation of power
output. The solar insolation variable is combined with one of the above mentioned
metrological variables and then power output of PV modules is determined. There
are various models in which various combination of temperature, pressure, cloud
variation, humidity, etc., are made with solar irradiance data (throughout the day)
of a particular geographical location, and used as input variables for power output
function.
The accuracy of this method depends on the accuracy of metrological data and
method of regression used. The simple and multiple regressions give better results
than single input (persistence model) method [16].
As the input data of solar irradiance and metrological have nonlinear nature,
nonlinear approaches of determination of relation or functions are used further for
better results. Artificial neural network is one of the approaches, which consider
wide range of nonlinear inputs and determined by different models. These models
are trained by real-input and -output data. Input data consist of more than one input
and activation function is selected on behalf of nature of nonlinearity of input and
output data. There are no assumptions made for training of models, therefore, they
produce better output than statistical methods.
In this method, there is input layer that accepts input data, hidden layers (which
may be more than one) determine the effect of input on output defined by weights
and bias. Output layer produces output data in the neural network model.
Initially, these weights are randomly selected for training of model and acti-
vation function squeezes sum of all signal and sent to output layer. For training,
reference input–output data are used. Output of model is compared with reference
output, further weights and bias are modified and again output is determined. The
weights are modified so as to reduce error between new output and reference output
than previous error. In this process, weights are modified each time to obtain better
output and it continues till the obtained output matches with reference output. On
completion of training, weights are obtained that validate reference input–output
data. After training of model, it can be used to obtain output from new set of
input data. There are various types of neural network layer interconnection or archi-
tecture, some of them are multilayer perceptron, multilayer feed forward neural
network, radial basis function neural network, recurrent neural network, adaptive
neuro-fuzzy inference systems (ANFIS), etc. These architectures have been used for
solar power output analysis. Auto-regressive integrated moving average (ARMA),
Support vector machine (SVM), ANFIS are frequently used for forecasting [15].
Some of the details of forecasting methods have been tabulated for the optimization
of power output/irradiance.
Above methods have been used for various inputs at different geographical loca-
tions to forecast hourly power output of PV modules. Machine learning refers to
methods to analyze real-time response to real-time changes in defined inputs, the
methods are normally combination of regression, neural, fuzzy, genetic, recognition
of weather input, etc., also reinforcement learning method. Reinforcement learning
normalize inputs and information is processed. Its decision-making system performs
action/evaluation and output is obtained, the environment (inputs and state variables)
respond on output produced. This information policy reward makes tuning in deci-
sion strategy/action to tune new output [17]. New output is a function of change in
input and state variables. This method may include above different machine learning
techniques to produce outputs constraint to environment limits.
5 Conclusion
As solar power is essential to be utilized for generation expansion, its intermittent

nature should be controlled to regulated output. Grid integration is controlled by
power converters, relevant algorithms are employed for gate signal of converters
to control voltage, frequency, etc. Accuracy in forecasting infers to reliability in
power market trade and less congestion conditions in transmission. As solar power
output is forecasted correctly, overall power market economics improves. This may
be achieved by machine learning algorithms while incorporating solar insolation,
weather data for short-term forecasting. There are various methods that have reduced
error in forecasted output to great extent. When neural, fuzzy and GA are combined
with one another and implemented to analyze the effect of weather on power output,
they have determined almost same output as real power produced in some cases.
SVM and reinforced learning along with previous algorithms are recent areas for
forecasting power output.
References
1. ErcanIzgi et.al (2012) Short–mid-term solar power prediction by using artificial neural networks
Solar Energy 86 (2012) 725–733
2. Kardakos EG et al (2013) Application of Time Series and Artificial Neural Network Models
inShort-term Forecasting of PV Power Generation Power Systems Laboratory, Department of
Electrical and Computer Engineering Aristotle UniversityofThessaloniki54124. Thessaloniki,
GREECE
3. A. Mellit et.al (2013) Artificial neural network-based model for estimating the produced power
of a photovoltaic module Renewable Energy 60 (2013)
4. Atsushi Yona,et.al(2013) Determination Method of Insolation Prediction WithFuzzy and
Applying Neural Network for Long-TermAhead PV Power Output Correction IEEE trans-
actions on sustainable energy, vol. 4, no. 2, April 2013
5. Almonacid F et al (2014) A methodology based on dynamic artificial neural network for short-
term forecasting of the power output of a PV generator. Energy Convers Manag 85(2014):389–
398
6. C. Cornaro et.al (2015) Master optimization process based on neural networks ensemble for
24-h solar irradiance forecast Solar Energy 111 (2015) 297–312
7. Francisco J.L. Lima et.al (2016) Forecast for surface solar irradiance at the Brazilian North-
easternregion using NWP model and artificial neural networks Renewable Energy 87 (2016)
807e818
8. MonowarHossain et.al (2017) Application of extreme learning machine for short term output
power forecasting of three grid-connected PV systems Journal of CleanerProduction 167 (2017)
395e405
9. A. Mellit, A. MassiPavan, V. Lughi(2019) Short-term forecasting of power productionin a
large-scale photovoltaic plant. Solar Energy, pages 402–413
10. http://www.ieapvps.org/fileadmin/dam/public/report/statistics/IEAPVPS_T1_35_Snapsho
t2019-Report.pdf
11. B. Wu, Y. Lang, N. Zargari, S. Kouro, —Power Conversion and Control of Wind Energy
System, John Wiley 2011,
12. G. Abad,.J. Lo´pez, M. A. Rodrı´guez, L. Marroy, —Doubly Fed Induction Machine, Modeling
and Control for Wind Energy Generation., John Wiley, 2011, pp 304, 481-483
13. Daniel S. Kirschen, Goran Strbac, ‘Fundamentals of Power System Economics’, John Wiley
& Sons 2004
14. Bhattacharya Kankar, Bollen Math, Daalder Jaap E (2001) Operation of Restructured Power
Systems. Springer, US
15. Utpal Kumar Das, Kok Soon Teya, Mehdi Seyedmahmoudiana, SaadMekhilef, Moh Yamani
Idna Idris, Willem Van Deventerc, Bend Horanc, Alex Stojcevski, ‘Forecasting of photovoltaic
power generation and model optimization: A review’
16. Oudjana SH, Hellal A, Mahamed IH. Short term photovoltaic power generation forecasting
using neural network. In: Proceedings of the 11th international conference on environment and
electrical engineering (EEEIC) 2012. IEEE; 2012.pp. 706–11
17. Mohssen Mohammed, Muhammad Badruddin Khan, EihabBashier Mohammed Bashier,
‘Machine Learning Algorithms and Applications’ CRC Press, Taylor & Francis Group, 2017
Chapter 26
Performance Comparison of Minimum
Shift Keying and Gaussian Minimum
Shift Keying in Additive White Gaussian
Noise Environment
Bablu Kumar Singh and Supriya Vyas
1 Introduction
Minimum Shift Keying (MSK) is a continuous phase modulation scheme. The modu-
lated carrier does not contain phase discontinuities and frequency changes at carrier
zero crossings. It is typical for MSK that the difference between the frequency of
logical 0’s (f0) and 1’s (f1) is equal to half the data rate. MSK modulation makes the
phase change linear and limited to ± (π/2) over the symbol interval. Due to the linear
phase change effect, better spectral efficiency is achieved. That means that MSK is
ordinary FSK with the modulation index set to 0.5 [1]. In an MSK, signal properties
can be observed are the envelope, is constant, phase is continuous between the bit
transitions and form of FSK signal, which involves two frequencies.
1.1 Gaussian Filter
The one-dimensional Gaussian filter has an impulse response expressed with standard
deviation as parameter:
1 −x 2
g(x) = √ · e 2σ 2 (1)
2π · σ
Here, σ is the standard deviation (Figs. 1 and 2).
B. K. Singh (B) · S. Vyas

M.B.M Engineering College, JNV University, Jodhpur, Rajasthan, India
e-mail: bablukumar.singh@gmail.com
S. Vyas
e-mail: supriyavyas77@gmail.com

https://doi.org/10.1007/978-981-15-8820-4_26
278 B. K. Singh and S. Vyas
Table 1 Parameters of MSK

Parameters for MSK Values set
MSK symbols 1000
Samples per symbol 16
Symbol rate (Hz) 400 k
Filter length (symbols) 8
Gaussian filtering is used to remove noise and details of a signal. Gaussian filters
have the properties of having no overshoot to a step function input while minimizing
the rise and fall time [2]. Graphically, we can see below a familiar bell-shaped
Gaussian distribution.
The Gaussian function defines the probability distribution for noise or data and it
is used for smoothing waveform by avoiding abrupt changes in the varying amplitude
of the signal.
Fig. 1 Gaussian filter

response
Fig. 2 MSK modulator

26 Performance Comparison of Minimum Shift Keying … 279
1.2 Raised Cosine Filter
This type of filter often used for pulse shaping in digital modulation due to its
skill to minimize inter-symbol interference (ISI). The frequency response |H(f)| of a
perfect/ideal raised cosine filter is symmetrical about the center of 0 Hz [3]. In our
application in MSK modulation and demodulation, the overall magnitude response
of the Nyquist filter is split between the transmitter and receiver. The phase response
compensates for the transmitting filter phase so to make it linear phase with following
equations below.
G(f) = GT (f)GR (f) (2)
√
|GT (f)| = |G R (f)| = |G(f)| (3)
∠|GR (f)| = −2π f τo− ∠GT (f) (4)
Here, GT (f) and GR (f) are the transfer functions of the transmitting and receiving
filters accordingly. The receiving filter in this case is matched to the transmitting filter
to maximize the signal-to-noise ratio (SNR) at the sampling time instances at the
receiver. The transmitting and receiving filters are typically considered to be linear
phase with a nominal delay of that is required to make the filters physically realizable
[4] (Figs. 3 and 4).
Fig. 3 MSK demodulator

Fig. 4 LabView code for MSK modulation and demodulation
2 MSK Modulator
The MSK modulation is a constant envelope signal with continuous phase that
results from modulating the instantaneous frequency with rectangular pulses. MSK
is considered to be a special case of Offset QPSK (OQPSK) with half sinusoidal
pulse weighting rather than rectangular [5]. The block schematic diagram of an
MSK modulator is shown in the figure.
The expression of MSK signal is
S M S K (t) = A cos[(2π f c t + φn (t)], nTb ≤ t ≤ (n + 1)Tb (5)
Here, A is amplitude of the carrier, fc is frequency of carrier, φn (t) which is

continuous is known as additional phase function. Tb is the symbol period.
3 MSK Demodulator
The MSK modulated signal can be demodulated either by a coherent or a differen-

tial (noncoherent) demodulation technique. The coherent demodulator scheme, for
QPSK and other modulations, needs precise reference of carrier, phase, and frequency
for optimum demodulation. This is not that easy for an MSK modulated signal.
Hence a suboptimal differential modulated technique along with threshold detection
is a popular choice in cellular telephony [1]. The figure of the MSK demodulator is
shown below.
Depending upon the value of the phase state θ(0), the inphase component equals
+g(t) or – g(t),where
⎧
⎨ 2Eb πt
cos , − Tb ≤ t ≤ Tb
g(t) = Tb 2Tb (6)
⎩
0, other wise
And depending upon the value of the phase state θ(Tb), the quadrature component
equals +g(t) or – g(t), where
⎧
⎨ 2Eb πt
cos , − Tb ≤ t ≤ Tb
g(t) = Tb 2Tb (7)
⎩
0, other wise
Probability of error/Bit error rate for coherent MSK is given as:

1 Eb
er f c (8)
2 No
And the bit error rate for coherent GMSK [5] is:

1 α Eb
er f c (9)
2 2No
4 Simulation of MSK in LabVIEW
MSK Modulator–Demodulator is designed in LabView and various parameter values

such as Samples per second is set to 16, symbol rate is kept 400 kHz and MSK symbol
is kept as 1000. The Eb/No value is then varied continuously and the corresponding
bit error rate is noted down first without using any filter and then using both Gaussian
and Raised cosine filters, respectively.
The theoretical results can be calculated using BER expression shown in Eqs. 8
and 9 to verify with the LabView results.
5 Results
A comparative analysis of simulated curves for BER vs. Eb/No (signal to noise ratio)
of MSK with and without filters is as shown below.
6 Conclusion
Performance comparison of Minimum Shift Keying’s demodulated signal using

Gaussian filter and raised cosine filter has been performed in this paper and the simu-
lated result as shown in Fig. 5, shows that with increase in Eb/No, BER decreases
and at lower Eb/No the performance of Gaussian filter is much better as compared
to raised cosine filter it can also be verified with constellation graph shown in Figs. 6
and 7, respectively.
Fig. 5 Performance analysis of BER vs Eb/No result analysis using LabView
Fig. 6 Constellation of MSK signal at 20 Eb/No without filter, Gaussian Filter and raised cosine
filter, respectively
Fig. 7 Constellation of MSK signal at 40 Eb/No without filter, Gaussian Filter and raised cosine
filter, respectively
References
1. Couch II LW (2001) Digital and analog communication systems, 6th Edn. Prentice-Hall, Inc
2. Cheng S, Iyer Sehshadri R, Valenti MC, Torrieri (2007) The capacity of noncoherent continuous-
phase frequency shift keying, in Proc. Conf. on Info. Sci. and Sys (CISS)
3. Tapkir PR, Singh SB, Thune NN (2016) Analysis and implementation of minimum shift keying
(MSK) modulation on FPGA platform. In: International Conference on Automatic Control and
Dynamic
4. Chen Y, Li Z, Liu P, Liu S (2017) The BER performance comparison of MSK and GMSK
schemes for short-range visible light communication. In: IEEE 9th International Conference on
Communication Software and Networks (ICCSN)”, Guangzhou, pp. 611–614
5. Shamee B, Christen L, Nuccio S, Yang J, Willner A (2010) Gausian Minimum Shift Keying
for spectrally efficient and dispersion tolerant optical communications,” CLEO/QELS: Laser
Science to Photonic Applications. San Jose, CA, pp 1–2
6. Molnar G, Babic M, Sipus Z (2018) Sharpened raised-cosine FIR filters, in 41st International
Convention on Information and Communication Technology, Electronics and Microelectronics
(MIPRO), Opatija, pp. 0111–0116
7. Symon H (2013) Digital communication Systems. Wiley 4th Ed
Chapter 27
MATLAB-Based Comparative Analysis
of Alternative PV Models
Khamma Kanwar and Jayashri Vajpai
1 Introduction
In the recent decades, Solar Photovoltaic (PV) systems have become the fastest-
growing alternative source of energy. These systems are environment friendly and
require less maintenance. Photovoltaic effect converts the solar radiation into the
electrical energy in solar cells. PV arrays have number of cells in series and parallel
or a group of panels. PV systems require little maintenance and are pollution free.
PV systems have standalone and grid connected applications such as, light sources,
battery charging, solar water pumping stations, hybrid vehicles, military and satellite
systems [1].
The nonlinear characteristic of PV cell can be modeled on the basis of Shockley
equation. Both single diode and double diode models are commonly used to simulate
PV characteristics. The single diode model simulates the PV cell characteristics
accurately. The efficiency of the solar energy conversion is directly related with the
maximum power extraction (MPP) from the PV system [2].
MPP is based on realizing the dynamic or static method of optimization. In the
dynamic method, the MPP is achieved by tracking of position and movement of the
sun. This approach may not be suitable for energy conversion at a small to medium
power range due to its high cost and energy consumption. In the static method
(for small power range), the MPP is tracked with the help of DC–DC converter by
adjusting continuously the operating point. In the literature, many methods have been
reported to track the MPP of PV generator. The Perturb and Observe (P&O) method
is widely used approach for tracking MPP [3].
K. Kanwar · J. Vajpai (B)

M. B. M. Engineering College, Jai Narain Vyas University, Jodhpur, India
e-mail: jvajpai@gmail.com
K. Kanwar
e-mail: kkjoya.jnvu@gmail.com

https://doi.org/10.1007/978-981-15-8820-4_27
286 K. Kanwar and J. Vajpai
Table 1 Comparison of datasheet parameter of three different PV systems

Mathematical model Inbuilt PV array User-defined PV array
Model KC200GT KC200GT KC200GT
Maximum Power [W] 200.143 200.143 200.143
Open circuit voltage Voc [V] 32.9 32.9 32.9
Short circuit current Isc [A] 8.21 8.21 8.21
Voltage at MPP Vmp [V] 26.3 26.3 26.3
Current at MPP Imp [A] 7.61 7.61 7.61
Diode saturation Current Io 9.825*10−8 2.3246e−10 8.9892e−16
[A]
Open-circuit −0.1230 −0.35502 −0.1230
voltage/temperature coefficient
KV [V/K]
Short-circuit 0.0032 0.06 0.0032
current/temperature coefficient
KI [A/K]
Ideality constant a 1.3 0.97736 0.64611
Shunt resistance Rp [] 415.405 150.6921 65.4647
Series resistance Rs [] 0.221 0.34483 0.47804
Number of cell in series Ns 54 54 54
Over the last decades, many researchers have studied the characteristics of PV
array by using simulation approach. Nguyen and Nguyen [4] have explained the step
by step procedure of mathematical modeling of PV array in MATLAB/SIMULINK
under different environment condition. Villalva et al. [5] have developed mathemat-
ical model to find out the best I–V equation for PV array by adjusting the I–V curve
at three points, viz., open circuit voltage (Voc), MPP and short circuit current (Isc).
Bouraiou et al. [6] have presented mathematical modeling of one diode and two
diode models of PV array and carried out simulation under different temperature and
irradiance conditions. The two algorithms have been compared experimentally by
Houssamo et al. [7] to maximize the output power from a PV system for the given
set of conditions using inbuilt PV array block in MATLAB. AbdelHady [8] has used
the inbuilt PV array and developed a MATLAB Simulink model that evaluates the
system performance when it is tied either with local grid or National grid. Some
of the researchers have used inbuilt PV array block and others have developed the
mathematical model of PV array. However, none of them has discussed which model
gives the best response.
In this paper, the mathematical model of PV array has been simulated in
MATLAB/SIMULINK on the basis of model equation. Then comparison of manu-
facture datasheet parameters and output response is made between the devel-
oped mathematical PV array model and inbuilt PV array block available in
MATLAB/SIMULINK in order to verify the accuracy of the inbuilt model.
27 MATLAB Based Comparative Analysis of Alternative PV Models 287
2 Methodology
The proposed comparative analysis of alternative PV models is carried out in the

following subsections.
2.1 Mathematical Modeling of PV Array
An overview of PV system is shown in Fig. 1 [3]. It consists of PV array along with

DC–DC Power converter that is connected to a PWM generator driven by MPPT
algorithm.
The four parameters model equivalent circuit diagram of PV array comprises of a
current source, a diode, a series resistance and a parallel resistance. The antiparallel
diode with the photo current source represents the theoretical model of ideal PV cell.
I–V characteristics may be expressed as follows [6]:

V + I Rs (V + I Rs )
I = I pv − I0 exp −1 − (1)
Vt a Rp
where
I pv Photocurrent (A)
I0 Diode saturation current (A)
Vt = Ns kT /q Thermal voltage of the array
Ns Number of cells connected in series
q Electron charge (1.60217646 *10−19 C)
k Boltzmann constant (1.3806503*10−23 J/K)
T Module operating temperature (K)
a Diode ideality constant
Rs Series resistance(ohm) and
Rp Parallel resistance (ohm)
Fig. 1 Overview of PV system [3]

The photocurrent of the PV cell depends linearly on the solar irradiation and
temperature according to the following equation:
G
I pv = (I pv,n + K I T ) (2)
Gn
where I pv,n photo current at the nominal condition generally 25 °C and 1000 W/m2 ,
T = T —Tn
Tn Nominal temperatures (K)
G Irradiation on the device surface (W/m2 ), and
G n Nominal irradiation
The diode saturation current is described by:
Isc,n + K I T
I0 = (3)
Voc,n +K V T
exp aVt
−1
Isc,n Short circuit current at nominal condition (A)

Voc,n Open circuit voltage at nominal condition (V)
KV Open circuit voltage/temperature coefficient KV [V/K]
KI Short circuit current/temperature coefficient KI [A/K]
Using these equations, the mathematical model of PV array has been developed
in MATLAB in many papers. The mathematical equation-based model of PV array
is simulated in this paper for the purpose of comparison.
2.2 Maximum Power Point Tracking
The I–V characteristics of PV array are nonlinear and show unique MPP under a
particular set of environment conditions, and this peak point keeps changing with
solar irradiation and ambient temperature. The efficiency of PV systems is generally
very low. For improvement of efficiency, the PV source impedance should match
with the load impedance for every operating condition, thereby requiring Maximum
power point tracking (MPPT) which optimizes the operating point [6].
The maximum power is set by adjusting the Duty Cycle of DC-DC converter. The
MPP can be tracked by various algorithms like:
a. Perturb and observe (P&O) method
b. Incremental Conductance method
c. Short circuit current method
d. Open circuit voltage method
e. Ripple correlation control method
P&O method is the most widely used MPPT technique on account of simplicity,
easy implementation and effectiveness.
Fig. 2 The flowchart of the P&O algorithm[8]
Perturb and Observe—In this approach, the voltage from the array is adjusted
in small bidirectional steps by the controller to observe the changes in the output
power. Further adjustments are done in the direction in which the power increases.
However, the P&O method can result in oscillations of power output due to the rise
and fall of the curve of power against voltage below the MPP [7]. The flowchart of
P&O algorithm is shown in Fig. 2 [8]
2.3 Modeling of DC/DC Converter
DC/DC converter is employed to convert a fixed voltage DC source into a variable-

voltage DC output. It provides high efficiency, good acceleration control and fast
dynamic response. There are mainly four basic topologies such as boost converter,
buck converter, buck-boost converter and Cuk converter [9]. However, since the
objective of this research work is comparison of models, the boost converter topology
has been employed for simplicity. The input and output voltage relationship can be
represented by following equation:
1
Vout = Vin
1− D
Here D is Duty Cycle, Vin is input voltage and Vout is output voltage.
3 Simulation in MATLAB/SIMULINK
The following table compares datasheet parameters of 200 W PV array model

KC200GT with MATLAB inbuilt PV array block and user-defined model that
have same reference parameters as per datasheet like maximum power, open circuit
voltage, short circuit current, voltage and current at MPP and number of cells in series
but remaining parameters are different. In user-defined model, there is provision to
define parameters but some parameters are fixed and some parameters can be defined
as per as datasheet. The mathematical model is developed in the preceding section
using the manufacturer datasheet parameters. It is expected that proper selection of
model may further improve the performance of model KC200GT.
3.1 Mathematical Modeling of PV Array

in MATLAB/SIMULINK
In this research work, MATLAB software has been used to implement the equation
of PV cell for developing a mathematical model of PV array KC200GT as shown
in Fig. 3. This model was designed for supplying maximum power to resistive load
with MPPT under particular environment conditions using Simulink blocks (Fig. 4).
Fig. 3 MATLAB/Simulink model developed by employing mathematical equations of PV system

Fig. 4 Subsystem of PV array
3.2 Inbuilt PV Array Block Available

in MATLAB/SIMULINK
The simulation was also carried out using the inbuilt PV array block model KC200GT
already available in MATLAB as shown in Fig. 5. The operating conditions were
designed for giving maximum power to the resistive load with MPPT. In this block,
some parameters are different from datasheet. Hence it is implemented by employing
Fig. 5 PV system using Inbuilt PV array block

Fig. 6 Output power characteristics of three different systems
the user-defined model. The performance of the developed mathematical model is

compared with the inbuilt PV array block model and user-defined PV models.
3.3 Simulation Results
The simulation is done for the three different systems as follows: In the first system,
the PV model is implemented using mathematical equations in Simulink. In the
second system, the PV model is implemented using inbuilt PV array block. In the
third system, PV model is implemented using user-defined block. The simulated
system output power characteristics are shown in Fig. 6.
It was found that the performance of PV model developed by using mathematical
equations is better than the system designed by inbuilt PV array model and user-
defined model.
4 Conclusions
The three different models of PV array KC200GT have been implemented in

MATLAB Simulink. Then comparative analysis of datasheet parameters of these
three different models of PV array has been carried out. The P&O algorithm is used for
giving the maximum power to the constant load with the help of boost converter. The
power characteristics obtained by mathematical equation-based modeling of PV array
gives better result than remaining two models. It is expected that this comparative
analysis will be useful for designing accurate models for research purpose.
References
1. Rahim KNA, Ping HW, Selvaraj J (2012) Photovoltaic module modeling using simulink/matlab.
In: The 3rd International Conference on Sustainable Future for Human Security SUSTAIN,
vol 17, pp 537–546. Elsevier (2013). https://doi.org/10.1016/j.proenv.2013.02.069
2. Surya Kumari J, Sai Babu Ch (2012) Mathematical Modeling and Simulation of Photovoltaic
Cell using Matlab-Simulink Environment. Int J Electr Computer Engineering (IJECE) 2(1):
26–34, ISSN 2088-8708
3. Jazayeri M, Uysal S, Jazayeri K (2014) Evaluation of Maximum Power Point Tracking Tech-
niques in PV Systems using MATLAB/Simulink, Sixth Annual IEEE Green Technologies
Conference, pp 54–60. IEEE computer society. https://doi.org/10.1109/GREENTECH.2014.21
4. Nguyen XH, Nguyen MP (2015) Mathematical modeling of photovoltaic cell/module/ arrays
with tags in Matlab/Simulink. https://doi.org/10.1186/s40068-015-0047-9
5. Villalva MG, Gazoli JR, Filho ER (2009) Comprehensive Approach to Modeling and Simula-
tion of Photovoltaic Arrays, IEEE TRANSACTIONS ON POWER ELECTRONICS, 24, NO.
5, pp. 1198-1208, 2009. https://doi.org/10.1109/tpel.2009.2013862
6. A. Bouraiou, M. Hamouda, A. Chaker, M. Sadok, M. Mostefaoui, S. Lachtar: Modeling and
Simulation of Photovoltaic Module and Array based on One and Two Diode Model Using
Matlab/Simulink
7. I. Houssamo, F. Locment, M. Sechilariu: Experimental analysis of impact of MPPT methods
on energy efficiency for photovoltaic power systems, Elsevier Electrical Power and Energy
Systems, VOL. 46 pp. 98–107, http://dx.doi.org/10.1016/j.ijepes.2012.10.048
8. R. AbdelHady: Modeling and simulation of a micro grid-connected solar PV system, science
direct water science, VOL. 31, pp. 1–10, 2017, http://dx.doi.org/10.1016/j.wsj.2017.04.001
9. Esram T, Chapman PL (2007) Comparison of Photovoltaic Array Maximum Power Point
Tracking Techniques. IEEE Trans Energy Convers 22(2):439–449. https://doi.org/10.1109/
TEC.2006.874230
10. A. Kumars, K. L. Ratnakar, B. S. Yogananda and B. R. Kamath.: Simulation of Photovoltaic
Cell and MPPT Controllers and their Analysis, International Journal of Engineering Science
and Innovative Technology (IJESIT) Vol. 4, pp. 253–263, 2015
11. Badii G, Ali H, Othman H, Rachid D (2014) Experimental Study of MPPT Algorithms for PV
Solar Water Pumping Applications. https://doi.org/10.5339/qfarc.2014.ITSP0107
12. Saoud MS, AbbassI HA, Kermiche S, Ouada M (2013) Improved incremental conductance
method for maximum power point tracking using cuk converter, WSEAS Trans Power Syst
8(3)
Chapter 28
Nuclear Fusion: Energy of Future
Jayesh Nehiwal, Harish Kumar Khyani, Shrawan Ram Patel,

and Chandershekhar Singh
1 Introduction
Fusion of light nuclei is by far the foremost process of energy release of energy in the
universe. The sun, like all of the stars, is nerved by the fusion of hydrogen nuclei into
one helium nucleus. The process of fusion releases the huge energy of 26.72 MeV
or 6.7 MeV per nucleon, which means around 180 MWh energy per gram. Hence,
the energy released in fusion reaction per unit mass of fuel is nearly seven times
the burning of coal. This efficaciously exemplifies the ultimate economic tackling of
matter which is chosen by nature for its prime energy release process. Further, these
and other fusion mechanisms are also liable for the generation of elements in the
stars. Nuclear fusion is a crucial and highly tempting option for a future terrestrial
supply of energy, since this source of energy would be environmentally agreeable
and the fuel comprising in it is practically limitless. Out of the nearly 80 plausible
fusion reactions, this reaction (Figs. 1, 2 and 3)
D + T → 4 He (3.517 MeV) + n (14.069 MeV)
between the isotopes of hydrogen, i.e., deuterium, D, and tritium, T, is of utmost
use. Around 20% of energy is carried away by alpha particle 4 He, whereas the
remaining 80% of energy is released as the kinetic energy of neutron n. The concen-
tration of Deuterium on Earth is practically limitless and is available in the form
of heavy water in the oceans which is 0.015% in natural hydrogen. Tritium does
not exist due to its radioactivity characteristics with a half-life of 12.35 years, so
for a reaction of D–T we have to generate Tritium through the breeding reaction of
Lithium
6
Li + n → T + 4 He + 4.8 MeV
7
Li + n → T + 4 He + n -2.47 MeV
J. Nehiwal (B) · H. K. Khyani · S. Ram Patel · C. Singh

Jodhpur Institute of Engineering & Technology, Jodhpur, India
e-mail: jayeshnehiwal@gmail.com

https://doi.org/10.1007/978-981-15-8820-4_28
296 J. Nehiwal et al.
Fig. 1 Average nuclear energy binding curve (Courtesy of Wikimedia Commons)
Fig. 2 Schematic diagram of Tokamak (Springer series in Material Science, vol. 214, pg. 314)
28 Nuclear Fusion: Energy of Future 297
Fig. 3 Flow chart of fusion reactor (reference [3])
where n is a slow neutron, which has a contribution for the generation of tritium.
Natural lithium comprises 7.42% of 6 Li and 92.58% of 7 Li. So we use deuterium
and lithium as fuel for fusion as they are cheap. According to the present scenario,
the resources of deuterium and lithium would be available for more than 30 million
years. The radioactive tritium is processed internally only in [1]. For the fusion of
two atoms, there is a requirement of binding energy. This binding energy is the
converse energy of mass defect according to Einstein’s theory of E = (m) c2 . This
is the energy required for disassembling the nucleus from an atom into separated
nucleons and the energy spent on separating each nucleon is average binding energy
[2]. For all such elements, the binding energy is shown in Fig. 1. From which it can
be interpreted that average binding energy for hydrogen isotopes and helium atom
are relatively low and hence fusion can be quite easily possible to meet the conditions
[2].
Under normal conditions of temperature and pressure, the fusion cannot take
place. For fusion to take place atoms are to be heated so that the motion of electrons
and nuclei will increase and electrons will get separated. The state of matter where
electrons and nuclei are not together is called a plasma state. The plasmas are very
hot and move with a very high velocity which makes the reaction very tough to
control as we have to overcome the high repulsive forces. So it is worth to gravitate
that fusion depends on plasma temperature and plasma density [3]. For the fusion
ignition, there is a requirement of the crossing energy value of the triple product
of temperature, density and energy confinement time. The triple product of D–T
reaction is two orders lesser than that of D–D & D–4 He reaction [3]. Hence it is
convenient to go for a fusion of D–T.
2 Confinement
For the fusion reaction, there is a requirement to control the plasma state, i.e., proviso
to confine the plasma, which can be done by mainly three processes. The first one
is gravitational confinement, which occurs between stars only. The second one is
magnetic confinement, which makes use of Lorentz force of magnetic field for the
confinement of charged particles. The third one is inertial confinement, in which
the fuel is made pellets to pressurize highly and then further the reaction is being
proceeded [3].
2.1 Magnetic Confinement
It is evident that no solid material can sustain heavy temperatures of 100 million
Kelvin so this dilemma is solved by stating the theory that all charged particles can
be confined under magnetic fields. It inspired to go for toroidal closed arrangement.
Only toroidal confinement also has limits to let escape some massive plasma so a
better technique is to provide superposition of strong toroidal field with poloidal field
[1]. The high temperature plasma experience Lorentz force, which is responsible for
confinement. Tokamak is a device or, more precisely, it is reactor which is of donut
shape surrounded by magnetic coils [4, 8, 9] as shown in Fig. 2.
2.2 Inertial Confinement
It is entirely different school of thought for fusion process. It compresses the fuel
into small solid pellets and due to extreme denseness when these pellets are heated
then the possibility exists for microexplosion. Altogether here the confinement is not
actually reached by external fields but the inertia binds the atoms together [3, 10].
Basically fuel pellet is enclosed by the layer of another material and extreme intense
beams of laser are subjected on the charged particle of high energy. The layer that is
the outer one gets heated up and evaporates. The products formed in evaporation get
mobile outward, and the pellet is compressed inward, due to momentum conservation
(inertial binding) [3]. The initiation of such fusion is done with the help of laser beam
and this can even also be done directly by laser driving the pellet itself and indirectly
by making use outer another layer usually gold layer and then laser driving.
3 Alternative Fusion: Muonic Fusion
This is an idea which is being investigated in few laboratories, to produce muons,

which are so called as the heavy sisters of the electron. The muon is used to inject
into a mixture of deuterium–tritium gas and there exist little probability that muon
will get captured by the tritium or the deuterium atom and there will be formation of
the deuterium–tritium molecule. Due to the heavy mass of muon, the dimensions will
be smaller as compared with normal molecule with bounded electrons [3]. Hence
the fusion reaction may undergo as the nuclei are closer to each other. But there is
problem to this as well, as the cost involved in production of muons is too much and
can catalyze fusion reactions about 200 times only [5].
4 Fusion Energy Conversion and Its Transfer
With the help of light isotopes, it is possible to have a fusion reaction possible, i.e.,
D–T fusion and this identification was responsible reason which has inspired to work
on various other factors to utilize this energy as a source of electrification. And hence
the need of design of such power plant is there. The features like steam generator,
turbine and current generator will be same as that in conventional fossil-fuelled or
nuclear power plants. A simple flowchart telling about the material and energy flow
can be depicted from Fig. 3. The fuel is injected and the conditions required for
fusion with system confinement are done which further leave the plasma and they
are stopped in the blankets which are the plasma surrounding modules [3].
5 Blanket System
This is the key system that converts fusion energy into electrical energy. Its main func-
tions are energy extraction, tritium breeding and radiation shielding, which means the
transferring of energy from fusion to thermal is being done, it ensures self-sufficiency
and it protects external subsystem from neutron and gamma radiation which may
cause an explosion. The heat which is produced in the blanket is transported via
water or helium to the steam generator and then it is fed to grid. A small part of
the same is utilized to power other components of plant. Cryo-system requires the
electrical power which produces low-temperature helium for the super-conducting
magnets, the current drive, the current in the magnets and the heating systems of
plasma [2].
6 India’s Role in Fusion Energy
R. Srinivasan and the Indian DEMO Team reviewed that the contribution to energy
from the fusion reactor is expected to rise beyond 2060. It will be beneficial if the
research program gets accelerated so that fusion reactors can be realized well before
2050 [6]. The multibillion project of ITER which was launched in 2007 to build a
Tokamak reactor and to check its feasibility, around 35 countries of the world took
their participation for the research and India is also part of that. Around 58% of the
ITER project has been completed and is been expected to generate power by 2035
[11].
7 Future Directions
Fusion energy is itself very unique in energy sector and challenging for R&D. ITER
is the biggest fusion reactor experiment that expects around 500 MW of power
generation in 400 s of pulse operation. The technology involving breeding capacity
of blanket is important and is still ongoing R&D but other factors of self-sufficiency
of tritium, economic, environmental friendliness, safety are also important [3, 7].
8 Conclusion
Fusion energy is safe and clean energy but needs efforts in implementing for the
applications. There is availability of various materials to fuse them like D, T, 3 He,
but need extreme conditions to initiate the reaction. The reactor for such fusion is still
under research and the issues are expected to be solved by ITER and DEMO by 2025.
Taming the fusion energy is a promising challenge but its rewards are incalculable
for the future energy needs of mankind.
References
1. Schumacher (2001) Status and problems of fusion reactor development”, Springer-Verlag,

ISSN 0028–1042, Volume 88, Issue 3, pp 102–112, March 2001
2. Yican W, Sumer S (2018) Fusion Energy Production. Elsevier- Comprehensive Energy Systems,
vol 3, pp 539–589
3. Hamacher T, Bradshaw AM (2001) Fusion as a Future Power Source: Recent Achievements
and Prospectus. World Energy Council, 18th Congress, Buenos Aires
4. Kikuchi M., Lackner K. Tran MQ Fusion Physics Vienna. IAEA 2012
5. Alvarez LW et al (1957) Phys Rev 105:1127–1128
6. Srinivasan R and the Indian DEMO Team (2010) Role of Fusion Energy in India J. Plasma
Fusion Res. SERIES, Vol. 9 (2010)
7. Mehlhorn TA (2014) National Security Research in Plasma Physics and Pulsed Power: Past,
Present, and Future. IEEE Transactions on Plasma Science, 42(5):1088–1117
8. Meade DM (1996) Recent Progress on the Tokamak Fusion Test Reactor. Journal of Fusion
Energy„ 15(3–4):163–167 ISSN 0164-0313
9. Cohn DR, Schwartz J, Bromberg L, Williams JEC (1988) Tokamak Reactor Concepts Using
High Temperature, High-Field Superconductors. J Fusion Energy 7(1):91–94
10. Ido S (1983) The Laser Fusion Reactor. J Fusion Energy, 3(5–6):459–463 ISSN 0164-0313
11. thehindubusinessline.com
Chapter 29
Medical Image Processing Using Soft
Computing Techniques
and Mathematical Morphology
Pratik Bhansali and Sandip Mehta
1 Introduction
Today, the mass information produced by medical imaging has considerably

increased. All the medical examinations carried out like for the preparation of neuro-
surgical operations or for the study of cerebral pathologies (cerebrovascular accident,
tumor, sclerosis in plate, etc.) can represent several hundreds of megabytes (100 Mo
= more than 0.8 billion bits 0 or 1). Physicians and cognitive scientists need reli-
able tools to assist them in their decision-making and in interpreting the mass of
information created.
The developments in magnetic resonance imaging (MRI) have allowed us to know
and evaluate the different pathologies of human beings with a degree of detail never
seen before. Many post-processing techniques have now been developed to efficiently
visualize large volumes of data and study the functionality of certain organs, such
as the brain, using the images generated by CT and MRI. Almost all postprocessing
tools require a spatial coherence of the data. If the acquisitions are multiple in times,
studying the same volume several times, it is necessary to ensure the spatial agreement
of the voxels by means of image processing techniques.
To interpret image is to produce symbolic description of it, that is, to recognize
and describe the different entities that make it up. Among the interpretation tools,
segmentation (or tagging) is a vital link in many applications and quantitative analyses
[1–4]. Segmentation is defined as the partitioning of an image into constituent regions
(also called classes or subsets), which are homogeneous with respect to one or more
P. Bhansali (B) · S. Mehta

Department of Electrical Engineering, Department of Electronics and Communication
Engineering, Jodhpur Institute of Engineering & Technology, Jodhpur (Rajasthan), India
e-mail: pratik.bhansali01@gmail.com
S. Mehta
e-mail: sandip.mehta@jietjodhpur.ac.in

https://doi.org/10.1007/978-981-15-8820-4_29
304 P. Bhansali and S. Mehta
Fig. 1 MRI acquisition (image a), tissue segmentation (image b), segmentation into structures
(image c) and segmentation of lesion (image d)
characteristics such as intensity, texture or functionality [5]. One of the typical use
cases of segmentation is the distinction of neuroanatomical zones, in which it is used
in order to separate areas of cerebral interest such as: hippocampus, frontal cortex,
areas of language, among others [6, 7].
The goal is, for one or more images, to give each voxel a label to give a description.
In cerebral MRI segmentation, we generally consider different level of descriptions
(see Fig. 1).
Tissue segmentation, which aims to describe the composition of each voxel among
the three main topics of the brain [8], the segmentation in structures, which describes
the membership of each voxel with a known brain structure: thalamus, putamen,
ventricular system, etc. and the segmentation of the cerebral lesions (tumors, multiple
sclerosis, cerebral vascular accidents) in the clinical setting, which aims to study their
location, volume, etc. Segmentation is, therefore, a central tool in both the clinical
and the neuroscience fields.
2 Segmentation in Medical Imaging
In medical imaging, it is necessary to take measurements on acquired images.

Whether anatomical or functional, the image provides accurate information about
patient such as: level of tracer uptake, volumes of different organs or regions, depth
of injury, etc. In many of these cases, the interest of the measure lies in a particular
area of the image associated with an anatomical region. In order to be able to make
these measurements, we use what we call segmentation techniques [9, 10].
Automatic image analysis and classification, in general, is based on the charac-
teristics of the image. These are basically two: the signal strength of the pixels and
their geometric arrangement in the image. Out of these characteristics, the inten-
sity is probably the most used, along with the information of contours that gives us
knowledge about the borders between the different objects of the image.
29 Medical Image Processing Using Soft Computing … 305
The techniques based on intensity base operation in the assumption that a type of
fabric has a characteristic and own brightness. This is generally not true, so it is not
possible to segment only by these techniques, requiring a postprocess to eliminate
regions with an intensity similar to the object of interest that do not belong to it [11].
In the literature on MRI segmentation, the methods are usually divided into two
categories: monospectral methods, if there is a single acquisition or modality of
image, and multispectral, if they use one or several images with different char-
acteristics but with geometric correspondence for the same anatomical volume
[12, 13].
The difficulty of segmenting an image is due to the structural complexity of the
MRI images and the contrast often insufficient to extract the structure of interest
without any knowledge of priori neither on its form nor on its location. Thus a
priori knowledge provided by the atlases probabilistic models will allow to better
differentiate close anatomical structures contrast and spatial location [14].
Many approaches have been proposed to segment cerebral MRIs. They can be
classified into two groups: methods supervised, requiring interaction with the user and
unsupervised methods, where the parameter estimation is performed automatically.
3 Fuzzy C-Means
Fuzzy C-Means (FCM) is a fuzzy unsupervised classification algorithm, it introduces

the basic concept of fuzzy set into class definition: every point in the dataset belongs
to each cluster with a certain degree, and all clusters are characterized by their center
of gravity. It uses a criterion of minimizing intraclass distances and maximizing
interclass distances, but giving a certain degree of membership to each class for each
pixel [15, 16].
This algorithm requires a prior knowledge of number of clusters and then generates
the classes through an iterative process by minimizing an objective function. Thus,
it makes possible to obtain a fuzzy partition of image by giving each pixel a degree
of belonging to a given class [17]. The cluster with which a pixel is associated is the
one with the highest degree of membership. The main steps of the Fuzzy C-means
algorithm are:
1. The arbitrary fixation of a membership matrix.
2. The computation of class centroids.
3. The readjustment of the membership matrix according to the position of the
centroids.
4. Calculation of the minimization criterion and return to Step 2 if there is no
convergence of criteria.
4 Fuzzy C-Means-Based Segmentation
Let X = {x1 , x2 , x3 , . . . x N } a set of N observations to classify in K classes where

each observation xi (i = 1, 2, 3, . . . ..N ) can be represented by a set of D attributes:
T
x j = x j1 , x j2 , x j3 , . . . x j D .
K Classes are represented by a vector of class centers
V = (v1 , v2 , v3 , . . . vk ), where vk = (vk1 , vk2 , . . . , xk D )T , is the center of class k.
Each observation xi is defined by its degree of belonging μik to the class k, that
μik ∈ [0, 1]. We can then define a partition matrix U = [μik ] of dimension (N ∗ K ).
The FCM algorithm consists in partitioning the N observations into K classes in
order to minimize the similarity of observations within each class. It results in the
minimization of the following objective function:

K
N
JFC M (U, V ) = μik
m 2
d (xi , vk ) (1)
k=1 i=1
Under constraints following:

N
K
0< μik < N , μik = 1 (2)
i=1 k=1
m: is the fuzzification factor or fuzzy factor such that 1 < m < ∞.

d(xi , vk ) = xi − vk is the distance between the observation i and the center of the
class k.
Generally the distance used is Euclidean:

D
2
d(xi , vk ) = xi j − vk j (3)
j=1
The first constraint ensures that no class should be empty and the second is a
normalization constraint which ensures that the sum of the degrees of membership
of each observation to all classes equal to 1.
Blurred partitioning is achieved by an iterative optimization of the objective func-
tion given by Eq. (1), with an update of membership levels and centers of classes as
in the case of the K-means algorithm.
Update of the following degrees of membership:
1
= (4)
K d(xi ,vk )
m−1
2
j=1 d(xi ,vk )
Hence the formula for updating class centers:

N
i=1 μik ∗ x i
m
vk = N (5)
i=1 μik
m
In what follows, we present the algorithm of the FCM method:

(1) Set the parameters: K: number of classes; ε: threshold representing the conver-
gence error (for example: = 0.001);m: degree of blur, usually taken as 2; Tmax: max
iteration; t = 1
(2) Initialize the matrix membership degrees 0 by random values in the interval
[0, 1]:
Calculate the matrix 0 by the relation (5)
Calculate the objective function 0 by the relation (1);
(3) Update the degree of membership matrix by the relation (4)
(4) Update the matrix (the centers) by the relation (5);
(5) Calculate objective function by the relation (1);
(6) Increment the value of t;
t = t +1 (6)
(7) Repeat steps 3–6 until you meet the stopping criterion which is:
(Jt − 1 − Jt ) < εout ≥ Jmax (7)
Finally, a final step is necessary when the desired result is a non-blurred. This
is called “defuzzification.” When we do not want to highlight the pixels where the
degrees of membership are approximately the same for each class, a natural way
to proceed is to consider that the final class of a pixel is that for which degree of
membership is maximal.
However, this algorithm that requires knowledge of the number of classes is not
robust and its effectiveness depends strongly of the initialization stage of the class
centers because the iterative process can easily provide a locally optimal solution.
Thus to improve the results of the classification, several methods are adopted for
optimizations of the result.
5 Particle Swarm Optimization
The clustering of images is a main task in the analysis and recognition of which
intend to divide pixels of an image into different classes such that the pixels of same
class are similar as possible and the pixels of diverse classes are also dissimilar as
possible. In spite of the efforts devoted to this problem leading to abundant literature,
it remains a very difficult task because of the lack of information, the distribution of
the pixels as well as a single and precise definition of a cluster.
PSO algorithms have shown their effectiveness in solving optimization problems

whose research space is large. Swarm particle optimization is a popular metaheuristic
approach based on swarm intelligence. The particle swarm optimization algorithm
launches the search with a population of solutions, where each is called “particle.”
The latter is characterized by a speed of movement and a position in the search space.
During the search process, each particle moves to change its position in the search
space based on its current speed, current position, and best position found in past
iterations, and best position found by the searcher “Swarm.” Its displacement allows
it to update its position and speed of movement at each iteration.
6 PSO-Based Segmentation
First, the performance function (fitness) is defined to start with our PSO-based
approach, which reflects the ability of individuals to adapt to the problem under
consideration. This function, which the algorithm seeks to minimize, is closely
dependent on the objective function.
Further, a particle is from the best overall solution and its best solution, the greater
the variation in its velocity, in order to move the particle towards the best solutions.
The random variables p1 and p2 can be defined as follows:
p1 = c1 * r1, p2 = c2 * r1.
With r1 and r2 following a uniform law on [0…1]. c1 and c2 are constants that
represent a positive acceleration, with c1 + c2 ≤ 4.
The algorithm runs as long as a convergence criterion has not been reached. This
may be:
• A fixed number of iterations.
• Depending on the fitness.
When the speed variation is close to 0.
The movement of the agents is controlled in terms of pbest and gbest. The speed
of agent i is updated first taking into account the current position ( pxi , pyi ) with
respect to its pbest:

vxi − rand() ∗ c1 i f pxi > pbest xi
vxi = (8)
vxi + rand() ∗ c2 i f pxi < pbest xi
Then it is updated with respect to gbest according to:

vxi − rand() ∗ c1 i f pxi > gbest x
vxi = (9)
vxi + rand() ∗ c2 i f pxi < gbest x
The parameters c1 and c2 will be replaced by constant 2 which produced better

results although they left as a future topic the investigation of their optimal value. In
the simplified version of these simulations (Current Simplified Version), the agents
updated their speeds by means of the equation:

vi, j = vi, j + 2 ∗ rand() ∗ pbesti, j − pi j + 2 ∗ rand() ∗ gbest j − pi j (3)
where the subscript i, j denoted agent i and dimension j.

In this work, the fuzzy segmentation is considered as common optimization
problem then PSO algorithm is used:
The objective function of each of the same function algorithm of the PSO method:

K
N
JFC M (U, V ) = μik
m 2
d (xi , vk ) (10)
k=1 i=1
7 Morphological Operation for Postprocessing
Morphological techniques probe the given image with a small template or shape
called structuring element (S.E.) [19]. The S.E. is sited at all possible locations in
the image and it is then compared with corresponding pixels of neighborhood.
After applying both FCM- and PSO-based algorithms for the purpose of
segmenting human brain MRI image morphological operations are applied to extract
the region of interest, i.e., to the removal of nonbrain region form the segmented
image.
Morphology process images with the help of previously chosen special shapes,
which are generally smaller than the image and are called the structural element,
which acts as an operator on an image to produce a result. The shape, size and
orientation of the structural element are chosen based on prior knowledge taking
into account the relevant geometric structures that are present in the image and the
objective pursued with the morphological operation implemented.
Each structure element requires the definition of a point of origin (or reference)
for its application as a morphological operator. This allows the structure element to
be related in a particular way to the pixels of the image.
After applying FCM- and PSO-based clustering segmentation technique, the clus-
tered information of the image achieved is further processed through morphological
operation to achieve the desired region of interest, i.e., the brain image from the
dataset of MRI chosen for segmentation.
First, an initial mask is created to filter the background of images with a threshold
and then the small regions of binary images are filtered. The attributes of the image
obtained are compared by the reference image to calculate the precision.
8 Results
The database is obtained from “Department of Telecommunications Brno University

of Technology” form signal processing laboratory [20]. This database contains a set
of actual medical images of MRI examination of several brain exams. There are
total 12 sets of training data available, i.e., MRI image sets of 12 different patients.
Each image set of a patient consists of 257 slices of image with resolution of 400
× 400 px and 16 bit grayscale bit depth. Each examination is composed with the
set of sequences of modalities, where each sequence constitutes all the images. The
number of sequences differs according to the examination.
The image sets contain not only original MRI images but also segmented labeled
image slices of each patient done by the human expert with the accuracy of 99.540% +
0.07756 which acts as are ground truth, comparing to which we validate the accuracy
of our segmentation methods. The simulation results on the random database images
of patient number 9 after applying both the proposed methodologies are presented
and then a comparative analysis on both visual and statistical ground is done to
validate the work.
To find the accuracy of our simulated result, we use the contingency matrix that
evaluates the performance of system by measuring the accuracy [21]:
TP +TN
Accuracy = (11)
T P + T N + FP + FN
A comparative visual analysis of both the methods, i.e., FCM- and PSO-based
segmentation is presented on Image slice number 70, taken from image dataset of
patient number 9 to showcase our methodology which can be implemented on all the
image slices of the dataset. Here the initially defined clusters are varied to judge the
preeminence of the proposed PSO-based approach for segmenting the brain image
MRI.
Image slice no.: ‘Image_70’:
Method- “Fuzzy C Means”
(a) Cluster = 3
The above shown images in Fig. 2 from image (a) to image (h) revels the
sequence of step by step appearance of results obtained as while performing
Fuzzy C means.
Here the cluster no. is initialized to be three (3) and the process for segmenting
the brain image is followed.
(b) Cluster = 4
The above shown images in Fig. 3, i.e., from image (a) to image (h) revel the
sequence of step by step appearance of the result when the cluster number is
initialized to be 4 and the process for segmenting the brain image is followed.
Fig. 2 Flow diagram for the

proposed work
(c) Cluster = 5
The process is repeated by initializing the cluster number to be 5. The images in
Fig. 4 from image (a) to image (h) revel the sequence of step by step appearance of
the results obtained.
Method—“Particle Swarm Optimization (PSO)”
(a) Cluster = 3
a) Input image b) Clustered image c) Initial Mask
d) Segmented Brain Mask e) Segmentation By Human Expert f) Segmentation Error
Fig. 3 Segmented brain image processed by FCM with cluster = 3

The above shown images in Fig. 5, i.e., from image (a) to image (h) revel the sequence
of step by step appearance of the results obtained while performing PSO.
Here the cluster number is initialized to be 3 and the process of segmenting the
brain image is followed.
(b) Cluster = 4
The process is repeated by initializing the cluster number to be The images in
Fig. 6, i.e., from image (a) to image (h) revel the sequence of step by step appearance
of results obtained.
(c) Cluster = 5
The process is repeated by initializing the cluster number to be 5 and the process
for segmenting the brain image is followed.
Table 1 shows the comparative analysis of the accuracy obtained after applying
both Fuzzy C Means and proposed PSO segmentation methods on different random
image slices of patient number 9 and its variation with respect to the change in the
initial cluster defined.
From the statistical evaluation shown in Table 1, a graph is plotted between cluster
number and accuracy which shows the comparative behavior of FCM- and PSO-based
segmentation approach on image no._70, i.e., “70” number slice of patient 9, which
was also visually evaluated in the previous segment. The same can be followed for
all the other image slices present in the dataset.
Fig. 6 Segmented brain image processed by PSO with cluster = 3

100 99.57
99.36 99.4
99.16
99
98
97.39
Accuracy
97
96 95.76
95
94
93
3 Cluster no. 4 5
(Image no._70)
FCM PSO
Fig. 9 Graphical Representation between cluster number and accuracy

Table 1 Result for patient number 9

S. no. Image number Cluster number Accuracy
FCM PSO
1 50 3 98.02 99.3
2 4 99.40 99.44
3 5 99.22 99.48
4 60 3 97.99 99.53
5 4 99.65 99.67
6 5 98.94 99.59
7 70 3 95.76 99.36
8 4 99.50 99.57
9 5 97.39 99.16
10 80 3 96.98 99.48
11 4 99.62 99.67
12 5 95.71 99.43
13 100 3 96.12 99.08
14 4 99.11 99.13
15 5 98.78 99.13
16 120 3 87.26 94.3
17 4 94.3 94.42
18 5 94.26 94.22
19 150 3 95.32 98.32
20 4 98.72 98.74
21 5 98.14 98.18
22 160 3 95.45 98.69
23 4 98.89 98.91
24 5 98.38 98.58
25 200 3 98.50 99.52
26 4 99.51 99.6
27 5 99.11 99.57
Conclusions and Future Scope of Work

The advancement in Image Processing techniques has opened a wide area of research
especially in the application involving medical imaging where accuracy in diagnosing
images plays a very important role. This paper has used two soft computing tech-
niques, which involve FCM and PSO method followed by mathematical morphology
to extract the region of interest from the brain MRI images.
The analysis on a patient’s human brain MRI image having 257 different image
slices has been done using MATLAB software. The preliminary results obtained
from the proposed method may be used to achieve better results in terms of accuracy
or segmented error of the MRI image. Further, it is evident from the study carried
out that the use of the proposed PSO algorithm allows increasing the performance
of the segmentation method under random variation of initial number of cluster. It
is found that more stable outcome was obtained in terms of accuracy as compared
to FCM method which gives unstable outcome with random variation of the initial
number of cluster.
The proposed method applied on MRI database images which may be explored
on different types of medical records like Ultrasonic images, X-Ray, CT scan, etc.
However, further research and development is needed to improve the existing tech-
nique and introduce new possible solutions that may preserve the sharp edges in the
template image.
References
1. Gonzalez RC, Woods RE (2002) Digital Image Processing. 3nd edition, Prentice Hall
2. Alasdair MC (2004) Introduction to Digital Image Processing with MATLAB. Cenage Learning
3. S Jayaraman, S Esakkirajan and T Veerakumar, “Digital Image Processing”, Tata McGraw
Hill, 2009
4. K. R. Castleman, “Digitial Image Processing”, Pearson, 1996
5. Rogowska, J., 2000. Overview and Fundamentals of Medical Image Segmentation-5
6. Worth AJ, Makris N, CavinessJr VS, Kennedy DN (1997) Neuroanatomical segmentation in
MRI: technological objectives. Int J Pattern Recognit Artif Intell 11(08):1161–1187
7. Despotović, I., Goossens, B. and Philips, W., “MRI segmentation of the human brain: chal-
lenges, methods, and applications”. Computational and mathematical methods in medicine,
2015
8. Yazdani, Sepideh, RubiyahYusof, AmirhoseinRiazi, and AlirezaKarimian. “Magnetic reso-
nance image tissue classification using an automatic method.” Diagnostic pathology 9, no. 1,
pp. 207, 2014
9. Pohle, R. and Toennies, K.D., “Segmentation of medical images using adaptive region grow-
ing”. In Medical Imaging (pp. 1337–1346). International Society for Optics and Photonics,
July 2001
10. Balafar MA, Ramli AR, Saripan MI, Mashohor S (2010) Review of brain MRI image
segmentation methods. Artif Intell Rev 33(3):261–274
11. Lladó, X., Oliver, A., Cabezas, M., Freixenet, J., Vilanova, J.C., Quiles, A., Valls, L., Ramió-
Torrentà, L. and Rovira, À.,” Segmentation of multiple sclerosis lesions in brain MRI: a review
of automated approaches”. Information Sciences, 186(1), pp. 164–185, 2012
12. Cabezas M, Oliver A, Lladó X, Freixenet J, Cuadra MB (2011) A review of atlas-based
segmentation for magnetic resonance brain images. Comput Methods Programs Biomed
104(3):e158–e177
13. Balafar, M.A, “ Fuzzy C-mean based brain MRI segmentation algorithms”. Artificial Intelli-
gence Review, pp. 1–9, 2014
14. Fedorov, Johnson, EswarDamaraju, Alexei Ozerin, Vince Calhoun and Sergey Plis. “End-to-end
learning of brain tissue segmentation from imperfect labeling” International Joint Conference
on Neural Networks (IJCNN), 2017
15. Bai Xiangzhi, Sun Chuxiong, Sun Changming (2019) Cell Segmentation Based On FOPSO
Combined With Shape Information Improved Intuitionistic FCM. IEEE Journal of Biomedical
and Health Informatics 23(1):449–459
16. Tao Lei, XiaohongJia, Yanning Zhang, Senior Member, IEEE, Lifeng He, Senior Member,
IEEE, Hongying. “Significantly Fast and Robust Fuzzy C-Means Clustering Algorithm Based
on Morphological Reconstruction and Membership Filtering” IEEE TRANSACTIONS ON

FUZZY SYSTEMS,vol- 26, no.- 5, pp: 3027 – 3041, 2018
17. Hiba Amin Mohammed Ali, Mohamed A.A. Ahmed, Eltahir Mohamed Hussein “MRI Brain
Tumor Segmentation Based on Multimodal Clustering and Level Set Method” International
Conference on Computer, Control, Electrical, and Electronics Engineering (ICCCEEE), 2018
18. Kennedy J (2011) Particle swarm optimization. In: Encyclopedia of machine learning, pp
760–766. Springer US
19. Reinoso O, Sebastián JM, Aracil R, Torres F (2001) Morphological operations with subpixel
resolution on digital images. Machine Graphics and Vision 10(1):89–102
20. Signal Processing Laboratory, Department of Telecommunications Brno University of Tech-
nology, 2011. http://splab.cz/en/research/konference-a-workshopy/challenge-2013/challenge-
3-brain-tissue-analysis
21. Robert S, Mustofa AA, Christy Atika Sari, De Rosal Ignatius Moses Setiadi, Eko Hari Rach-
mawanto (2018) MRI Image Segmentation using Morphological Enhancement and Noise
Removal based on Fuzzy C-means.5th International Conference on Information Technology,
Computer, and Electrical Engineering (ICITACEE)
Chapter 30
Analysis of Hotspot Development
in Power Transformer and Its Life
Estimation
Vinit Mehta and Jayashri Vajpai
1 Introduction
The operational efficiency and the economic viability of power systems are governed
largely by the functioning and the cost of its constituent power transformers. The
reliability of power systems is adversely affected by the failure and maloperation of
power transformers. These occur due to the failure of insulation caused by high stress
under abnormal or critical operating conditions or in cases when heat generated in a
power transformer is not dissipated efficiently by the surrounding medium. Generally,
hotspots are developed in the power transformers when the heat dissipation is not
uniform or effective, leading to thermal stress.
This paper presents a technique for estimating the loss of life of transformer
by modeling the thermal stresses that are responsible for the deterioration of their
quality and performance and employing them to calculate accelerated aging. The
most important factor among these is the hotspot temperature (HST), which is a
major reason for the loss of life of transformer. The HST of a transformer primarily
depends on the ambient temperature, the rise in the top oil temperature (TOT) over
the ambient temperature and the rise in the winding HST over the top oil temperature.
HST values for different load conditions can be estimated by considering appropriate
computational model on the basis of the thermal characteristics of the transformer
and the cooling system.
This paper proposes a computational model that has been simulated on
MATLAB/Simulink and that has provision for evaluation of the HST for every hour
in a given load cycle. This is employed to estimate the aging acceleration factor. The
percentage loss of life is predicted on the basis of these values. Further, by providing
Oil Natural Air Forced (ONAF) cooling arrangement during peak load period, the
V. Mehta (B) · J. Vajpai

Electrical Engineering Department, J.N.V. University, Jodhpur, Rajasthan, India
e-mail: vinit741@gmail.com

https://doi.org/10.1007/978-981-15-8820-4_30
320 V. Mehta and J. Vajpai
saving in percentage loss of life is determined. The proposed model has been used
to predict the loss of life of a 315MVA power transformer in operation at 400 kV
GSS, Surpura, Jodhpur (Rajasthan, India). After Introduction, the paper includes
five more sections, that present the state of art, proposed methodology, algorithm,
MATLAB/Simulink model, results and discussion.
2 State of Art
In order to overcome the abnormal operating conditions and to increase the trans-
former loading capacity, different calculation procedures for estimating the winding
hotspot temperature with reference to load changes have been proposed by many
authors.
The base of their work is primarily IEEE or IEC guidelines. IEEE Guide for
Loading Mineral-Oil-Immersed Transformers [1] is applicable to loading mineral oil-
immersed distribution and power transformers, with different types of construction,
along with special considerations for the degree of conservatism involved in the
loading. This has paved the way for understanding and developing models for the
simulation of thermal characteristics of power transformers as attempted in this paper.
Hashmi et al. [2] have performed the steady-state calculations using IEC guidelines
to determine the hotspot temperatures of distribution and power transformers in the
worst environment due to long summer periods. Amoda et al. [3] have presented an
investigation into the adequacy of the IEEE HST model, when the model parameters
are to be determined from measured field data.
Shiyou et al. [4] have analyzed the mechanism of thermogenesis and thermolysis
of transformer along with the position of the hotspot temperature. Further, the calcu-
lation of the loss of insulation life of dry-type transformer has been carried out on the
basis of HST. Silva and Bastos [5] have addressed the influence of simplifications
to be made on the geometries of power transformers for the performance of thermo-
dynamic simulations to diminish the computational time and to obtain the magnetic
fields, temperatures and heat flow in the interior of the transformer. Humayun et al.
[6] have proposed the demand response and dynamic thermal rating-based optimiza-
tion model for efficient capacity utilization and life management of transformers
during contingencies. Kweon et al. [7] have estimated the hotspot temperature in the
power transformer by the optical fiber sensors and validated it by the conventional
heat run test. Gouda et al. [8] have introduced the HST- and TOT-based thermal
model under linear and nonlinear loads. Vanegas and Mahajan [9] have determined
the thermal characteristics, load profiles and acceleration factor equation of an oil-
immersed current transformer and compared the estimated and expected values of
the aging acceleration factor.
DejanSuja et al. [10] have presented an accurate temperature calculation method
based on the thermal-electrical analogy that considers nonlinear thermal resistances
at different locations within a power transformer. Yong Liang [11] has developed a
30 Analysis of Hot Spot Development in Power Transformer 321
graphical tool for predicting TOT using a semiphysical model to assess the effect of
solar radiation and wind velocity on the prediction of TOT.
Muhamad et al. [12] have investigated the effect of HST on the overall transformer
temperature distribution and its effect on the heat dissipation to the surface of trans-
former tank, for condition monitoring purposes. Further, simulation of mineral oil-
filled distribution transformer (ONAN type) has been done by using Finite Element
Method Magnetism software. Longnv et al. [13] have presented an accurate compu-
tational alternative for hot-spot temperature-rise estimations in a single-phase auto-
transformer to compute the stray losses using finite-element method (FEM), along
with the average surface convection heat transfer coefficients of the structural parts
of the transformer. Radakovic et al. [14] have developed a detailed thermal-hydraulic
network model for determining the value of the hotspot factor and the HST using
FEM. Srinivasan [15] has proposed a semiphysical model comprising of variable
environmental conditions for the estimation of HST in transformer and along with
a MATLAB/Simulink-based valid model. Takami et al. [16] have done an online
monitoring of the transformer using FEMLAB and MATLAB software to estimate
the HST of oil-immersed power transformers.
In general, many simplifying assumptions have been made in the various proposed
methods for calculating the HST of power transformers, as reported in the standards
documentation and published literature. The aim of this paper is to develop a compre-
hensive computational methodology for thermal model of power transformer, for
estimating accelerated aging factor and loss of life, along with ease of application
offered by the MATLAB software.
3 Proposed Methodology
This section presents the salient features of the proposed thermal modeling of a
three-phase power transformer by using MATLAB software and uses it to estimate
the loss of life of the power transformer. The theoretical aspects of the estimation of
loss of life are first described, followed by the algorithm.
3.1 Thermal Modeling of Top Oil Temperature Rise
The rise of top oil temperature over ambient temperature is an indication of contin-
uous loading of transformer. An increase in the load increases the losses thus
increasing the overall temperature. The rate of change of temperature depends upon
the overall thermal time constant of the transformer, which in turn depends upon the
heat capacity of the transformer, i.e., the mass of the core, coils, and oil, and the rate
of heat exchange from the transformer. The change of top oil temperature is modeled
as a first-order differential equation as follows [10].
dθTO (t)
TTO = θTO (u) − θTO (t) (1)
dt
where, TTO is the top oil time constant in minutes,θTO (t) is the top oil temperature
rise over ambient temperature in °C, θTO (u) is the final top oil temperature rise in
°C, and t is the time referenced to the time of the loading change.
Equation (1) is solved to obtain the following exponential response from the initial
temperature state to the final temperature state [10],

θTO (t) = [θTO (u) − θTO (i)] 1 − e−t/TTO + θTO (i) (2)
where, θTO (i) is the initial top oil temperature rise in °C.
The final rise in the top oil temperature depends upon the load factor and can be
approximated by the following equation:
n
K2 R + 1
θTO (u) = θTO (r) (3)
R+1
where, θTO (r) is the full load top oil temperature rise over ambient temperature
in °C, R is the ratio of load loss at rated load to no-load loss, K is the ratio of the
specified load to rated load, n is an empirically derived exponent that depends upon
the cooling method. The IEEE loading guide [1] recommends the use of n = 0.8 for
natural convection and n = 0.9 to 1.0 for forced cooling. The top oil time constant
at the considered load is given by the following:
Cth−oil ∗ θTO (r)

TTO = 60 ∗ (4)
qtot
where, qtot is the total supplied losses in W, and Cth−oil is the equivalent thermal
capacitance of the transformer oil in W-h/°C.
The equivalent thermal capacitance of the transformer oil is given by the following
equation:
Cth−oil = 0.48 ∗ MOil (5)
where, MOil is the weight of the oil in kg.
3.2 Thermal Modeling of Hotspot Temperature Rise
The increase in the transformer current due to losses increases the oil and
winding temperature. The change of hotspot temperature is modeled as a first-order
differential equation shown in Eq. (6) [10]:
dθHS (t)
THS = θHS (u) − θHS (t) (6)
dt
where, THS is the hotspot time constant in minutes, θHS (t) is the hotspot temperature
rise over top oil temperature rise in °C, θHS (u) is the final hotspot temperature rise
in °C and t is the time referenced to the time of the loading change.
This can be solved to obtain

θHS (t) = [θHS (u) − θHS (i)] 1 − e−t/THS + θHS (i) (7)
where, θHS (i) is the initial hotspot temperature rise in °C.

Based on the IEEE model, the final rise in the hotspot temperature considering
the load factor can be obtained by the following equation:
θHS (u) = θHS (r)[K]2m (8)
where, θHS (r) is the rated hotspot temperature rise over top oil temperature and m
is an empirically derived exponent that depends on the cooling method.
The winding hotspot time constant can be calculated as follows:
θHS (r)
THS = 2.75 ∗ (9)
(1 + Pe ) ∗ S2
where, THS is the winding hotspot time constant in minutes at the rated load, Pe is
the relative eddy current losses (W), S is the current density in A/mm2 at rated load.
Finally, the hotspot temperature is calculated by adding the ambient temperature,
the top oil temperature rise over ambient, and the hotspot temperature rise over top
oil. This can be expressed by the following equation [1].
θH = θA + θHS (t) + θTO (t) (10)
where, θA is the ambient temperature in °C and θH is the ultimate hotspot temperature

in °C.
3.3 Estimation of Equivalent Aging Factor
In oil-immersed transformers, paper or cellulose material along with oil forms the
major insulation. Therefore, the insulation must maintain adequate dielectric strength
against voltage surges and adequate mechanical strength against short-circuit forces.
As cellulose ages thermally in an operating transformer; three mechanisms
contribute to its degradation, namely; hydrolysis, oxidation, and pyrolysis. The agents
responsible for the respective mechanisms are water, oxygen, and heat. Each of these
agents will have an effect on degradation rate so they must be individually controlled.
Water and oxygen content of the insulation can be controlled by the transformer oil
preservation system but control of heat is left to transformer operating personnel.
Transformer insulation life is defined as the total time period between the initial
state for which the transformer insulation is considered new and the final state for
which dielectric stress or short circuit stress could occur in normal service and cause
an electrical failure.
Experimental evidence indicates that the relation of insulation deterioration to
time and temperature follows an adaptation of the Arrhenius reaction rate theory that
has the following form [1]:

B
Per unit life = A*exp (11)
θH + 273
where, A is a modified per unit constant and B is the aging rate. The temperature of
110 °C is selected for one per unit life.
The Aging Acceleration Factor (FAA ) per unit transformer insulation life is given
by the following equation [1].

15000 15000
FAA = exp + (12)
383 θH + 273
The equivalent loss of life (in hours or days) at the reference temperature in a
given time period for the given temperature cycle is given as follows.
N
FAAn ∗ tn
FEQA = n=1
N (13)
n=1 tn
where, FEQA is the equivalent aging factor for the total time period, n is the index
of the time interval, N is the total number of time intervals and FAAn is the aging
acceleration factor for the temperature which exists during the time interval tn .
3.4 Estimation of Percentage Loss of Life
The insulation per unit life curve is used to calculate the percent loss of total life
of a transformer. The normal insulation life at the reference temperature is defined
in hours or years. The values of normal insulation life for a well-dried, oxygen-free
system are given in Table 1.
The percentage loss of life is given as follows:
FEQA ∗ t*100
Percentage Loss of Life = (14)
Normal Insulation Life
Table 1 Normal Insulation Life of a Well-Dried, Oxygen-Free 65 °C Average Winding Temperature

Rise Insulation System at the Reference Temperature of 110 °C [1]
Basis Normal insulation Life
hours years
50% retained tensile strength of insulation 65 000 7.42
25% retained tensile strength of insulation 135 000 15.41
200 retained degree of polymerization in insulation 150 000 17.12
Interpretation of distribution Transformer functional life test data 180 000 20.55
Further, by providing ONAF cooling arrangement during peak load period, the
saving in percentage loss of life is determined. This can then be employed for the
evaluation of residual life.
The algorithm for thermal modeling of the power transformer is now presented.
4 Algorithm
Step 1 Initialize the input variables of transformer on hourly basis. This includes load
factor K, total losses q (W), ambient temperature θA (°C), OTI reading θTO (i)(°C)
and WTI reading θHS (i) (°C) every hour in a given load cycle.
Step 2 The rated values of rated top oil rise over ambient temperature θTO (r)(° C),
rated hotspot rise over ambient temperature θHS (r) (°C), exponent n & m, weight
of oil Moil (kg), current density S (A/mm2 ), relative eddy current loss Pe (W), ratio
of load loss at rated load to no-load loss R.
Step 3 The different values of final top oil temperature rise θTO (u)(°C) are
determined by placing the values of K, R and n on hourly basis, using Eq. (3).
Step 4 The different values of final hotspot temperature rise θHS (u)(°C ) are
determined by placing the values of K and m on hourly basis, using Eq. (8).
Step 5 The top oil time constant TTO (mins) is determined with the help of Eq. (4).
Step 6 The hotspot time constant THS (mins) is determined with the help of Eq. (9).
Step 7 The top oil temperature rise θTO (°C) is obtained from the application of
Eq. (2).
Step 8 The hotspot temperature rise θHS (°C) is obtained from the application
of Eq. (7).
Step 9 The hotspot temperature θH during a day on hourly basis is calculated
using Eq. (10).
Step 10 The aging acceleration factor is obtained using Eq. (12).
Step 11 The equivalent aging of the power transformer is calculated by using
Eq. (13).
Step 12 The percentage loss of life is obtained from the application of Eq. (14).
Ambient
Input
Ioad Theta_TO
Hot Spot
Top Oil Model

Load
Input Load
Load Theta_HS
Hot Spot Model
Fig. 1 Block diagram showing the thermal dynamic model of a power transformer
All the steps illustrated above, may be repeated for a load cycle that contains
overload conditions during an hour for a day. The percentage loss of life thus calcu-
lated shows the amount of loss of life of transformer that can be used to estimate the
reduced loss of life of the power transformer.
5 Matlab/Simulink Model
Figure 1 shows a simplified block diagram of the MATLAB/Simulink thermal

dynamic model of a power transformer.
Equations (2) and (7) are solved using MATLAB program for determination of the
top oil temperature rise θTO (t) and hotspot temperature rise θHS (t), respectively.
At each discrete time interval of 60 min, the top oil temperature rise and the hotspot
winding temperature rise are calculated.
The hotspot temperature θH is the sum of ambient temperature, top oil temperature
rise and hotspot temperature rise. The measured hotspot temperature θH results for
a 315 MVA, 400/33 kV transformer during the given load cycle are then used to
determine the residual life of the transformer.
In order to validate the proposed model, data gathered under various load condi-
tions from a real power transformer (315MVA) which are recorded in the month
of January, have been used. In this study, work has been carried out in a power
transformer situated at 400 kV GSS, Surpura, Jodhpur substation. The specifica-
tions, cooling arrangements and temperature measuring equipments of the proposed
power transformer are as shown in Tables 2, 3 and 4, respectively.
Table 2 Specification of a
Rated voltage (HV) 400 kV
315 mva 400 kV/33 kV power
transformer Rated voltage (LV) 33 kV
Rated current (HV) 454.70 A
Rated current (LV) 1837.00 A
Current Density 0.28A/mm2
No. of phase 3
Frequency 50 Hz
Connection Symbol Yd11
Weight of core and coil 129400 kg
Weight of tank and fittings 32850 kg
Weight of Oil 64090 kg
Rated top oil rise over ambient temperature 45 °C
Rated hotspot rise over top oil temperature 55 °C
Ratio of load loss at rated load to no-load loss 2
Exponent ‘n’ 0.8
Exponent ‘m’ 0.9
No load loss 17500 W
Relative winding eddy current losses 152 W
Table 3 Cooling Equipment

Oil pumps and fans Pump (600 gpm) Fan (467 cum per
used in 315MVA
capacity min)
400 kV/33 kV power
transformer No. of oil pumps and 4 10
fans (2 + 2) (8 + 2)
(Running +
Standby)
Table 4 OTI and WTI

OTI Alarm 95 °C
auxiliary contacts settings
Trip 100 °C
WTI Fan Start 85 °C
Pump Start 95 °C
Alarm 115 °C
Trip 125 °C
The thermal behavior of the power transformer has been evaluated and verified
using the top oil and hotspot temperature models. Results of these thermal models for
a power transformer are discussed in the following section. The typical load factor,
total losses, ambient temperature, OTI and WTI reading for a 315MVA, 400/33 kV
power transformer located at 400 kV GSS, Jodhpur are shown in Table 5.
Table 5 Input data to a 315 MVA transformer

Clock Load factor (K) Total losses Ambient OTI reading WTI
time (Watts) temperature (°C) reading (°C)
(°C)
6:00 am 0.57 685263 12 22 23
7:00 am 0.59 685905 13 25 24
8:00 am 0.61 696532 14 26 25
9:00 am 0.71 745623 14 24 26
10:00 am 0.78 795623 15 25 26
11:00 am 0.79 797010 15 29 29
12:00 pm 0.87 800100 17 28 29
01:00 pm 0.89 812356 18 27 30
02:00 pm 0.96 889654 18 26 31
03:00 pm 1.08 895698 18 25 32
04:00 pm 1.01 891258 18 30 32
05:00 pm 0.99 889932 17 31 31
06:00 pm 0.98 890008 15 32 30
07:00 pm 1.04 895498 14 35 30
08:00 pm 0.87 801124 14 36 28
09:00 pm 0.85 790050 14 25 27
10:00 pm 0.65 724613 13 26 24
11:00 pm 0.64 734212 13 29 23
12:00 am 0.55 686541 13 30 23
01:00 am 0.56 686689 12 32 22
02:00 am 0.57 689365 12 31 24
03:00 am 0.59 690635 11 35 25
04:00 am 0.61 696432 11 36 26
05:00 am 0.54 685252 12 35 24
The normal load profile of the 315MVA, 400/33 kV power transformer for 24 h in
winter season (January) is shown in Fig. 2. It is clear from Fig. 2, that the maximum
value of the load factor is 1.08 occurring at 03:00 pm.
The HST for every hour is evaluated by using a MATLAB program. The effect of
heat dissipated due to various losses, i.e., constant and variable losses in a transformer
on the useful life of a cellulose insulation material has first been estimated on a per
unit basis. Cumulative loss of life has been calculated for varying load conditions with
the understanding that one real day of operation will produce less or more aging than
one day. Figure 3 shows the graphical representation of the hotspot temperature for
a day including the variation of the ambient temperature, the top oil temperature rise
over ambient, and the hotspot temperature rise over top oil and the results are shown
1.2
1.1
1
Load Factor
0.9
0.8
0.7
0.6
0.5
0 5 10 15 20 25
time in hours
Fig. 2 Normal load cycle profile
120
100 θ
cummulative age hours
80
∆θ
60
∆θ
40
θ
20
0
0 5 10 15 20 25
time in hours
Fig. 3 Graphical representation of hotspot temperature, hotspot temperature rise, top oil tempera-
ture rise and ambient temperature
in Table 6. The graphical representation of transformer insulation life throughout a

day is shown in Fig. 4.
Now, a peak load having load factor of 1.5 is applied to the same transformer for
an hour during a day results in a rise of hotspot temperature. Again, the cumulative
loss of life is predicted, which will produce aging greater than 24 h. Table 7 shows the
result of loss of life due to the impact of peak load on the transformer. The graphical
representation of transformer insulation life when a peak load having load factor of
1.5 is applied for an hour during the same day is shown in Fig. 5.
The normal insulation life at the reference temperature in hours is assumed to
be 180000. The percentage loss of life is evaluated for different load conditions as
shown in Table 7.
It has been observed that employing ONAF cooling arrangement during peak
load period reduces the hotspot temperature to 135 °C. The cumulative loss of life is
Table 6 Analytical results for a 315 MVA transformer

Clock time θTO θHS θH FAA Aging Hours Cumulative
age hours
6:00 am 10.889 19.824 42.713 0.434 0.434 0.434
7:00 am 26.374 21.299 60.673 0.560 0.560 0.994
8:00 am 27.249 22.612 63.862 0.584 0.584 1.579
9:00 am 27.598 29.659 71.257 0.643 0.643 2.222
10:00 am 29.514 35.088 79.602 0.713 0.713 2.936
11:00 am 31.944 35.922 82.866 0.741 0.741 3.678
12:00 pm 32.856 42.686 92.543 0.829 0.829 4.507
01:00 pm 32.742 44.467 95.210 0.854 0.854 5.362
02:00 pm 34.111 50.931 103.042 0.930 0.930 6.292
03:00 pm 36.410 62.904 117.314 1.076 1.076 7.368
04:00 pm 37.349 55.788 111.137 1.011 1.011 8.380
05:00 pm 37.412 53.816 108.228 0.982 0.982 9.362
06:00 pm 37.713 52.838 105.551 0.955 0.955 10.317
07:00 pm 40.691 58.774 113.465 1.035 1.035 11.352
08:00 pm 37.346 42.678 94.024 0.843 0.843 12.196
09:00 pm 30.743 40.929 85.673 0.766 0.766 12.962
10:00 pm 27.834 25.316 66.151 0.602 0.602 13.565
11:00 pm 29.477 24.617 67.094 0.610 0.610 14.175
12:00 am 28.937 18.787 60.724 0.560 0.560 14.736
01:00 am 30.273 19.391 61.664 0.568 0.568 15.304
02:00 am 29.780 20.030 61.810 0.569 0.569 15.873
03:00 am 32.453 21.308 64.761 0.591 0.591 16.465
04:00 am 33.296 22.621 66.917 0.608 0.608 17.074
05:00 am 31.871 18.191 62.062 0.571 0.571 17.645
25
Age corresponding to
20
unity load factor
15
10
giiven operating load factor
0
0 5 10 15 20 25
time in hours
Fig. 4 Transformer insulation life
Table 7 Peak Load Applied to 315MVA Transformer

Clock Time Load Factor θH Faa Aging hours Cumulative
(K) Age Hours
6:00 am 0.57 42.713 0.434 0.434 0.434
7:00 am 0.59 60.673 0.560 0.560 0.994
8:00 am 0.61 63.862 0.584 0.584 1.579
9:00 am 0.71 71.257 0.643 0.643 2.222
10:00 am 0.78 79.602 0.713 0.713 2.936
11:00 am 0.79 82.866 0.741 0.741 3.678
12:00 pm 0.87 92.543 0.829 0.829 4.507
01:00 pm 0.89 95.210 0.854 0.854 5.362
02:00 pm 0.96 103.042 0.930 0.930 6.292
04:00 pm 1.01 111.137 1.011 1.011 27.793
05:00 pm 0.99 108.228 0.98 0.982 28.775
06:00 pm 0.98 105.551 0.95 0.955 29.730
07:00 pm 1.04 113.465 1.035 1.035 30.765
(continued)
Table 7 (continued)
Clock Time Load Factor θH Faa Aging hours Cumulative
(K) Age Hours
08:00 pm 0.87 94.024 0.843 0.843 31.609
09:00 pm 0.85 85.673 0.766 0.766 32.375
10:00 pm 0.65 66.151 0.602 0.602 32.978
11:00 pm 0.64 67.094 0.610 0.610 33.588
12:00 am 0.55 60.724 0.560 0.560 34.149
01:00 am 0.56 61.664 0.568 0.568 34.717
02:00 am 0.57 61.810 0.569 0.569 35.286
03:00 am 0.59 64.761 0.591 0.591 35.878
04:00 am 0.61 66.917 0.608 0.608 36.487
05:00 am 0.54 62.062 0.571 0.571 37.058
40
35 Age corresponding to
peak load for 1 hour
30
25 unity load factor
20
15
10
giiven operating load factor
5
0
0 5 10 15 20 25
time in hours
Fig. 5 Transformer insulation life corresponding to peak load for 1 h
now calculated again. This results in the saving in percentage loss of life as shown
in Table 8. The results obtained from thermal model may be used to estimate the
residual life of the power transformer.
The proposed model gives the approximate HST values that are in close agree-
ment with the measured field data. It can be concluded that with the increase in
the transformer temperature beyond thermal limits would reduce its life below the
specified normal life.
Table 8 Cooling Equipment

S. Condition of Operation % loss of life
Used In 315mva
No.
400 kV/33 kV Power
Transformer 1 Operating Load Factor every hour 0.0098
2 Peak load factor for an hour 0.0206
3 ONAF cooling during peak hour period 0.0153
Further, the study of insulation ageing is important, as it reduces both the mechan-
ical and dielectric-withstand strength of the transformer. An ageing transformer
is subjected to faults that result in high radial and compressive forces. Also, the
conductor insulation gets deteriorated and becomes unable to sustain the mechanical
stresses caused by a fault. Hence, it is the dominant factor in limiting the lifetime of
the transformer. Providing proper cooling such as ONAF has great potential to save
the percentage loss of life.
The thermal model proposed in this paper is dependent on the accuracy of esti-
mated steady-state temperature rises. Therefore, it will be important to develop a
variable time interval calculation method with estimation at smaller time intervals
when there are dynamic changes in temperature and larger intervals when the steady
state is achieved. However, further research and development are needed to improve
the existing monitoring systems and introduce designs and applications that include
better thermal modeling. These thermal models will allow the transformer manu-
facturers to provide better specifications and users to operate the transformers on
appropriate loading by considering the ambient temperature conditions.
References
1. Guide for Loading Mineral-Oil- Immersed Transformers, IEEE Standard C57. 1991–1995
2. Murtaza H, Matti L, Seppo H (2013) Effect of Climate Change on Transformers Loading
Conditions in the Future Smart Grid Environment. Open Journal of Applied Sciences 3:24–29.
https://doi.org/10.4236/ojapps.2013.32b005. Available:http://www.scirp.org/journal/ojapps
3. Oluwaseun AA, Tylavsky DJ, McCulla GA, Knuth WA (2008) A New Model for Predicting
Hottest-Spot Temperature in Transformers. IEEE Trans. Power Symposium, NAPS ‘08. 40th
North American, pp. 1–8. https://doi.org/10.1109/naps.2008.5307407
4. ShiyouWANG,Youyuan WANG and Xuetong ZHAO, “Calculating Model of Insulation Life
Loss of Dry-Type Transformer Based on the Hot-Spot Temperature”, IEEE 11th International
Conference on the Properties and Applications of Dielectric Materials, pp. 720-723, July 2015.
https://doi.org/10.1109/icpadm.2015.7295373
5. Juliano R. da Silva and Joao P. A. Bastos, “Analysis of Power Transformer Geometry Simpli-
fications on Electromagnetic and Thermodynamic Simulations”, IEEE Trans. on magnetics,
vol. 51, no. 3, article 8400404, March, 2015. https://doi.org/10.1109/tmag.2014.2358993
6. Muhammad Humayun, Mubbashir Ali, Amir Safdarian, Merkebu Z. Degefa, and MattiLe-
htonen, “Optimal Use of Demand Response for Lifesaving and Efficient Capacity Utilization
of Power Transformers during Contingencies”, IEEE Trans. Power & Energy Society General
Meeting, pp. 1 - 5, 2015. DOI: https://doi.org/10.1109/pesgm.2015.7285627
7. Dong-JinKweon, Kyo-Sun Koo, Jung-Wook Woo and Joo-SikKwak, “A Study on the Hot Spot
Temperature in 154 kV Power Transformers”, Journal of Electrical Engineering & Technology,
vol. 7, no. 3, pp. 312-319, 2012. Available:http://dx.doi.org/10.5370/JEET.2012.7.3.312
8. O.E. Gouda, G.M. Amer, W.A.A. Salem, “Predicting transformer temperature rise and loss
of life in the presence of harmonic load currents”, Ain Shams Engineering Journal, vol. 3,
pp. 113–121, 2012. Available: www.elsevier.com/locate/asej
9. Diego M. RobalinoVanegas and Satish M. Mahajan, “Correlation between Hot-Spot Tempera-
ture and Aging Factor of Oil-Immersed Current Transformers”, IEEE Trans. Power and Energy
Society General Meeting - Conversion and Delivery of Electrical Energy in the 21st Century,
pp. 1-5, 2008. DOI: https://doi.org/10.1109/pes.2008.4595981
10. Dejan Susa, MattiLehtonen, and HasseNordman, “Dynamic Thermal Modelling of Power
Transformers”, IEEE Trans. on power delivery, vol. 20, no. 1, pp. 197-204, January 2005
11. Yong Liang, “Simulation of Top-Oil Temperature for Transformers”, Thesis and Project Report,
Power Systems Engineering Research Center Cornell University, New York, February 2001.
Available:www.pserc.wisc.edu
12. N. A. Muhamad, H. Kamarden and N. A. Othman, “Heat Distribution Pattern of Oil-filled Trans-
former at Different Hottest Spot Temperature Locations”, IEEE 11th International Conference
on the Properties and Applications of Dielectric Materials, pp. 979–982, July 2015. https://doi.
org/10.1109/icpadm.2015.7295438
13. Longnv Li, ShuangxiaNiu, S. L. Ho, W. N. Fu and Yan Li, “A Novel Approach to Investigate
the Hot-Spot Temperature Rise in Power Transformers”, IEEE Trans. on magnetics, vol. 51,
no. 3, article 8400204, March, 2015. https://doi.org/10.1109/tmag.2014.2359956
14. Zoran Radakovic, UrosRadoman, and PredragKostic, “Decomposition of the Hot-Spot Factor”,
IEEE Trans. on power delivery, vol. 30, no. 1, pp. 403 – 411, Feb, 2015. https://doi.org/10.
1109/tpwrd.2014.2352039
15. M.Srinivasan and A. Krishnan, “Effects of Environmental Factors in Transformer’s Insulation
Life”, WSEAS transactions on power systems, vol 8, Issue 1, Jan 2013
16. Kourosh Mousavi Takami, Hasan Gholnejadand JafarMahmoudi, “Thermal and hot spot eval-
uations on oil immersed power Transformers by FEMLAB and MATLAB software’s”, IEEE
Trans. Thermal, Mechanical and Multi-Physics Simulation Experiments in Microelectronics
and Micro-Systems, pp. 1–6, 2007. https://doi.org/10.1109/esime.2007.359924
Chapter 31
Photovoltaic Module Cleaning Prediction
Using Deep Neural Networks
Kapil Panwar , Aditya Jindal , and Kusum Lata Agarwal
1 Introduction
With increase in population and per capita electrical consumption, world is quickly
moving toward renewable sources for energy generation. Solar and wind are the
most used renewable sources for generating electrical power. Solar photovoltaic
power plants are increasing in numbers and most of them are in plain grounds or
near to desert areas so as to obtain high irradiation profile for high yields. Due to
plain ground areas, soil accumulation on PV module is a common problem. This
soil accumulation on PV modules affects the generation. This creates a thin layer of
hindrance for irradiation and resulting generates lower electrical power [1].
To mitigate generation loss due to soil accumulation over PV modules, PV
modules need to be cleaned/washed. For cleaning of PV modules, required water
should be of drinking water type, with normal pH (pH 6–7) so as to avoid any harm
to its glass and its antireflection coating [2]. Hence treated water is used to clean PV
modules.
As most of the solar PV plants are in remote area as well as near to desert area,
there is already water scarcity. Due to water scarcity, modules cannot be washed on
daily basis using water. On the other hand, if borewell water is being used along with
water treatment plant then waste management is another challenge.
K. Panwar (B) · A. Jindal

Mahindra Teqo Pvt. Ltd, Bengaluru, India
e-mail: panwar.kapil2@mahindra.com
A. Jindal
e-mail: jindal.aditya2@mahindra.com
K. L. Agarwal
Electrical Department, JIET Jodhpur, Mogra Khurd, India

https://doi.org/10.1007/978-981-15-8820-4_31
336 K. Panwar et al.
Cleaning of solar PV module should be done in such a way that it minimizes water
consumption and keep plant in condition to generate optimum electrical power.
In this paper, artificial intelligence (deep neural network) technique has been
used to develop a model that predicts when solar PV modules should be cleaned,
and also which modules should be cleaned. As soil accumulation is different for near
pathways, clusters and switch yard due to vehicle movement and totally different in
middle of power plant [3]. By this water can be used wisely in cleaning only those
modules which are contributing a lot in generation minimization, instead of cleaning
whole plant.
2 Deep Neural Network
Artificial intelligence is a branch of science in which machines are made intelligent

like humans, to take their own decision. AI is a broadly used term, It has three popular
subbranches as which are shown in Fig. 1.
To make machines think like human, artificial brain is developed using artificial
neurons. These artificial neurons work in same ways as human neuron works.
ArƟficial
intelligence
Reinforcement
ConvenƟonal ML Deep Learning
Learning
Fig. 1 Branches of artificial intelligence
Fig. 2 Deep Neural Network [DNN]

31 Photovoltaic Module Cleaning Prediction Using Deep Neural Networks 337
Deep learning is a subbranch of AI in which a number of neurons are being

used with multiple hidden layers to make successive learning more accurate. These
algorithms require high-processing units like GPUs or TPUs to process millions of
data and train the model. Deep Neural Networks are used for big data problems,
computer vision and in solving complex problems.
3 Methodology
For this model creation, we selected one Solar PV plant located in the western region,
India. In that solar PV, plant SMA make central inverter is installed along with Trina
make PV modules.
3.1 Dataset Used for Model
In plant, there are lots of sensors available that measures weather parameters and DC
current, voltage and power of Inverter. This dataset is of 1-min resolution and stored
in local DB using SCADA. Hence 5 years of historical data are being used as input
to this model. Parameters used are as per the below list:
• Global Horizontal Irradiation (w/m2 )
• Ambient Temperature (°C)
• Rain Fall (in mm)
• Humidity (%)
• Inclined Irradiation (w/m2 )
• PV Module surface temperature (°C)
• DC Current (A)
• DC Voltage (V)
• DC Power (kW).
3.2 Impact of Change in Irradiation on DC Current and DC

Voltage
Using feature engineering, it is found that, change in irradiation directly impact on

DC Current and DC Power, whereas change in temperature has an impact on DC
voltage [4]. Sample IV curve has been shown in Fig. 3 and PV curve in Fig. 4.
Fig. 3 IV curve of solar PV module [Generic]
Fig. 4 PV curve of solar PV module [Generic]

Fig. 5 Impact of irradiation on avg DC current is linear
Fig. 6 Architecture of DNN model
3.3 Creating Model and Training
For solving this kind of problems, various kinds of models can be developed using
machine learning techniques such as SVM, logistic regression and anomaly detection,
etc. In this paper, deep neural networks are being used which is an advance technique
of artificial intelligence branch. Inspired from human neurons, algorithmic neurons
are developed and same as human brain, number of neurons are interconnected in
layers to develop a deep neural network (Figs. 5 and 6).
3.4 Neural Network Architecture
Neural networks are made of layers of neurons. For this model, three sections of
neuron layers are developed: Input layer, hidden layer and output layer. For the
same, nine neuron input layers and two neuron output layers are being developed.
3.5 Model Training and Optimization
To build this model, Python language is being used. Neurons are developed using
Ten-sorFlow framework. Training is the most critical part of any model. Due to huge
amount of data, “mini batch” technique is being used to train this model. Developed
model used SoftMax in last hidden layer so as to predict binary output for given set
of parameters.
To optimize model “Auto Encoder” technique is being used as this technique
optimizes time cycle as well as reduces memory consumption while training this
model [5]. This technique reduces dimension of weights which are being updated in
each learning cycle. This method is logically same as PCA which is used to reduce
dimensionality for any machine learning model.
4 Result
To solve this problem, the generated model is being trained using 80% of historical
data. To validate the accuracy of model, rest of the 20% data are fed to developed
model and compared predicted versus actual cleaning daily data for a specific inverter.
These validation data were selected arbitrary which were not part of training data so
as to avoid biasness.
These predicted data and actual data of validation set give more clarity. Figure 7
represents the predicted versus actual module cleaning results.
This result represents that cleaning date predicted by model was 88% correct
with actual cleaning dates whereas 12% times cleaning suggested dates were not
matched with validation cleaning set. Similarly, noncleaning dates were predicted
90% accurately.
5 Conclusion
This model can predict when a set of modules need to be cleaned with 88% accuracy.
This accuracy can be increased by tuning hyperparameter and including feedback in
every new training cycle.
Fig. 7 Result of model with validation set
There is a future scope for training this model with multiple site datasets so as
to make it generalized. Same model can be developed using the anomaly technique
and can be compared with the existing model performance.
Acknowledgements We provide our special thanks to Mahindra Teqo Pvt. Ltd. for providing real
data and opportunity to develop this model.
References
1. Wable SS (2017) Design and manufacturing of solar panels cleaning system. Int J Res Appl Sci
Eng Technol (IJRASET) 5(VII)
2. Khadka, N (2018) Solar panel cleaner technology: a review. In: Conference: 5th international
conference on developments in renewable energy technology (ICDRET’18), At Kathmandu
3. John JJ (2015) Evaluation and prediction of soiling loss on PV modules with artificially deposited
dust. In: IEEE 42nd Photovoltaic Specialist Conference (PVSC)
4. Sobha Rani P (2018) Effect of Temperature and Irradiance on Solar Module Performance. IOSR
J Electr Electron Eng (IOSR-JEEE) 13(2) Ver. III:36–40
5. Kiarashinejad Y (2019) Deep learning approach based on dimensionality reduction for designing
electromagnetic nanostructures. ARXIS J
Chapter 32
Optimal Controller Design for DC–DC
Buck Converter
Shubham Sharma and Kusum Lata Agarwal
1 Introduction
A DC–DC converter is used to convert one level of DC voltage to another. A linear

DC–DC converter uses resistive voltage drop to regulate output voltage but this
method is not efficient due to large power loss. To avoid this switch mode, DC–
DC converters are used. Switch mode DC–DC converter uses power electronics
devices and energy storing element which gives regulated dc output voltage and
higher efficiency [1].
The switch-mode power electronics converters are some of the simplest power
electronics converter which converts one level of electrical voltage into another level
by switching action. These converters have received an increasing deal of interest in
many areas. This is due to their wide application like power supplies for personal
computers, office equipment, appliances control, telecommunication equipment, DC
motor drives, automotive, aircraft, etc. They are also used in the satellites where DC
buses at different voltage levels are supplied through these DC–DC converters [1].
The DC–DC converter is having low cost, small size and high efficiency but it
contains large ripple in output and output voltage regulation is not good as compared
with linear power supply so a good controller is required to improve these perfor-
mance parameters. PID (proportional-integral-derivative controller) controller can
be used to improve output voltage regulation and ripple in output. The Pulse width
modulation (PWM) technique is used to generate the switching pulse [2]. The PID
controller generates a signal that is given to pulse width modulator. Pulse width
S. Sharma (B)
JIET-JODHPUR, Mogra Khurd, Rajasthan, India
e-mail: shubhamee81@gmail.com
K. L. Agarwal
Department of Electrical Engineering, JIET-JODHPUR, Mogra Khurd, Rajasthan, India

https://doi.org/10.1007/978-981-15-8820-4_32
344 S. Sharma and K. L. Agarwal
Fig. 1 Block diagram of

DC–DC converter with PID
controller
modulator adjusts the width of switching pulse such that the desired output voltage
is achieved. The block diagram of DC–DC converter with PID controller is shown
in Fig. 1.
The DC–DC converter controller can be designed for voltage mode control or
current mode control. The DC–DC converter is a nonlinear converter so design of
controller having lots of complexities. To design controller for nonlinear DC–DC
converter linearized model of converter is required. Most important thing in design of
a PID controller is the estimation of k p , ki , and kd (Tuning of PID controller) for best
performance of DC–DC converter [3]. In general, Trial and Error, Zeigler–Nichols
and Internal mode control (IMC) methods are used for tuning of PID controller.
This method of tuning does not give the optimal operation of PID controller. For
optimization of PID controller, (LQR) method is used for tuning of PID controller.
LQR controller estimates k p , ki , and kd to minimize cost function [2]. LQR controller
adjusts weight matrices to get the desired performance parameters. The weight
matrices can be determined using trial and error method, algebraic solution method
and genetic algorithm etc. That means LQR tuning method tunes PID controller to
minimize ripple in output and to improve output voltage regulation with minimum
cost function. The LQR method also suppresses the disturbance occurring in the
system with better performance parameters [2]. LQG controller is combination of
LQR controller together with Linear quadratic Estimator (LQE) for good voltage
regulation in output, for the rejection of small disturbances and for the elimination of
noise. In this method output measurements are assumed to be disturbed by Gaussian
noise.
The paper is organized as follows. In the second section, we have discussed briefly
about the DC–DC buck converter and way to obtain the linearized state-space model
of buck converter. In the third section, the PID tuning methodology using Zeigler-
Nichols, LQR and LQG methods is discussed. Simulation results are discussed in
the fourth section. Simulation is done in MATLAB Simulink environment.
32 Optimal Controller Design for DC–DC Buck Converter 345
2 DC–DC Buck Converter
DC–DC buck or step down regulator produces lower average output voltage as
compared with input voltage. It is used in power supplies and speed control of motor.
A DC–DC buck converter with resistive load is shown in Fig. 2.
DC–DC buck converter is low cost, small size and highly efficient converter but
its output contains ripple and voltage regulation is poor as compared with linear
converter. To improve ripple factor and voltage regulation, controller is required.
There are many control schemes available. The operation of buck converter can be
studied in two modes. Mode 1 means switch is in on position and mode 2 means switch
is in off position. Dynamics of DC–DC converter can be understood by state-space
representation of buck converter [3, 4].
Mode-1 Operation:
ẋ = A1 x + B1 u (1)
y = C 1 x + D1 u (2)

0 −1 1
where A1 = L
1 −1 , B1 = L , C1 = 0 1 and D1 = 0
C RC
0
Mode-2 Operation:
ẋ = A2 x + B2 u (3)
y = C 2 x + D2 u (4)
Fig. 2 DC–DC buck converter


0 −1
0
where A2 = , B2 =
L
1 −1 , C 2 = 0 1 , D2 = 0
C RC
0
From above, the average model of buck converter can be represented as
ẋ = Ax + Bu (5)
y = C x + Du (6)

0 −1
D/L
where A = L
1 −1,B= ,C = 01 , D= 0
C RC
0
Using average model, transfer function of DC–DC buck converter can be written
as:
Vd
G(S) = (7)
LC S 2 + (L/R)S + 1
3 PID Controller
PID controller performs proportional, integral and derivative action on error signal to
get desired response. PID controllers are widely used in the industrial control appli-
cations. Error signal is basically a difference between the real value and desired value
[5]. The most important task in the design of PID controller is to find proportional
gain kp , integral gain ki and derivative gain kd . The process of determining values of
these gains is known is the tuning of PID controller [6]. The PID controller improves
transient characteristics such as peak overshoot, rise time, settling time and reduces
steady-state error [7]. The tuning methods used are explained below.
3.1 Zeigler–Nichols Method
Zeigler–Nichols is the simplest method of tuning PID controller. Z–N method is a

heuristic tuning method that attempts to produce good values of kp , ki , and kd . To
tune PID controller using Z–N method, following steps need to be followed:
1. Make ki and kd zero until we get stable oscillation in the output. A gain at which
stable oscillations in the output occur are known as control gain ku .
2. After getting stable, oscillation determines the time period of one oscillation
which is termed as Tu .
3. After measuring ku and Tu , we can measure the PID controller gains kp , ki , and
kd are calculated by formula proposed by Zeigler–Nichols.
3.2 Linear Quadratic Regulator Method
The optimal control theory is the backbone of the modern control theory [8]. LQR is
an optimal control method of tuning PID controller [3, 9]. LQR problem is defined
as:
To Derive the State X of Linear Systems Ẋ = AX + BU to the Origin by
Minimizing the Following Quadratic Performance Index
∞
1
J= (X T Q X + U T RU )dt (8)
2
0
Using closed-loop state feedback gain:
U (t) = −K .X (t) (9)
where Q and R matrices are positive definite matrices. Matrices Q and R express
relation between energy expense and error rate [10]. The LQR method gives the
optimal value of weight matrices by solving algebraic Riccati equation.
A T P + P A − P B R −1 B T P + Q = 0 (10)
The LQR weight matrices Q and R have been calculated from the following
methods:
Trial and Error Method: In this method, we randomly select the values Q and R and
determine k p , ki , and kd . By doing some reasoning, right combination of matrices Q
and R is obtained but this method is very time consuming [3, 4, 9].
Algebraic approach: The implementation of LQR controller requires minimization
of cost function which is achieved by placing weights on control input and state
variables. It is assumed that:
R=r (11)
⎡ ⎤
q11 0 0
Q = ⎣ 0 q22 0 ⎦ (12)
0 0 q33
On solving Riccat equation, following values of parameter of q matrix were

obtained
r
q11 = 2
[2ζ ωn3 − A231 ] (13)
B31
r
q22 = 2
2[A33 + 3ζ ωn ]A31 − [A232 − (2ζ 2 + 1)2 ωn4 ] − [A31 + ζ ωn3 ]6ζ ωn (14)
B31
r
q33 = 2
[−2(A32 + (2ζ 2 + 1)ωn2 ) − (A233 − (3ζ ω)2 )] (15)
B31
So element of matrix Q can be obtained from above equations but for that we need
to define damping ratio ζ and natural frequency ωn . Steps followed for algebraic
approach are:
1. Design mathematical model of system to be controlled.
2. Now define damping ratio ζ and natural frequency ωn .
3. Obtain state gain matrix K in relation to p using ARE.
4. Now actual characteristic equation of system is determined.
5. Compare ARE with characteristics equation to obtain three elements of matrix p.
6. Fix the value of R and by using three elements of matrix p we need to determine
values of elements of matrix Q.
7. Now calculate gain matrix K.
8. Obtain response in MATLAB simulink.
Genetic Algorithm Method: GA is a useful and power tool to solve the optimization
problem with constrained and unconstrained condition. GA is basically based on
natural selection process of biological evolution. GA plays an important role in
optimal control problem. In this method, we find the values of gain matrices of LQR
for different values of Q and R by optimizing cost function J. The genetic Algorithm
problem is solved using following steps:
1. Initialization: The GA problem is solved by generating initial population.
The population contains individuals. The individuals are basically set of
chromosomes and each individual is having separate set of chromosome.
2. Fitness Function: The fitness function selection is very important procedure
of genetic algorithm. A function which is needed to be optimally minimized is
known as fitness function. In LQR problem, our cost function J is selected as
fitness function.
3. Selection: Now fit chromosomes are selected from the population two create off-
springs. This procedure is followed until we get chromosomes close to each other
and less diversity is obtained. For selection, roulette wheel selection method is
used.
4. Crossover: Crossover means reproduction. In this procedure, two parent chro-
mosomes will crossover to create new child chromosome.
5. Mutation: As we know that in biological sense, child characteristics are not same
as parents characteristics. There are some changes in genes at the time of growth
of children which makes child different from parents. In GA, we change some
bits of chromosomes to change characteristics of child this is known as mutation.
Algorithm for GA method for LQR problem is:
1. Start
2. The initial population of K is selected randomly.
3. Select the weight values on state variable and control input i.e. Q and R
4. Calculate the fitness function of each chromosome. If optimal J is obtained than
go to step 7.
5. Now the reproduce next generation.
6. Change the bits of next generated chromosomes by applying crossover and
mutation operations.
7. Determine best possible optimal value of gain matrix for given Q and R.
8. If result is not satisfying than go to step 3 otherwise go to next step.
9. End.
3.3 Linear Quadratic Gaussian
LQG controller is basically combination of LQR controller and Kalman filter (known
as linear quadratic state estimator). LQG controller is applicable to linear system only
and system is driven by the white Gaussian noise. So basically to design an LQG
controller, we need to solve LQE problem than LQR problem. To design LQE, we
assume that system is disturbed by noise. Our model is represented as:
ẋ(t) = Ax(t) + Bu(t) + ω (16)
y = C x(t) + Du(t) + v (17)
In this method, state X(t) is recovered by information known to us as output Y
and control signal U. LQE tries to minimize the estimated error e(t) = x(t) − x (t)

where x(t) is unknown real state vector and x (t) is a estimated vector. Our main aim
is to minimize the cost function:
Je = E[eest
T
(t).eest (t)] (18)
Now estimator will solve equation above and provide solution as

[u(t)
x (t) = (A − LC)x (t) + B L (19)
y(t)]
where L = PC T V −1 .
Table 1 Comparison of transient poerfomance

S.No Controller Rise time Settling time Peak overshoot Steady-state error
(ms) (s) (%) (V)
1 Z-N 160.969 0.018 26.52 0.02
2 LQR-TE 3.747 0.0025 26 0
3 LQR-AA 3.225 0.00030 22.72 0.08
4 LQR-GA 3.869 0.00034 26 0.09
5 LQG-TE 3.274 0.003 26 0
6 LQG-AA 26.39 0.024 25 −0.05
7 LQG-GA 3.281 0.00028 8 0
The tuning of PID controller was proposed in the previous sections. The conventional
approach of designing a PID controller (Zeigler-Nichols) has been implemented
but controller was not able to provide better transient performance and not able
to compensate the change in input voltage and load. The LQR controller provided
better transient performance and able to compensate the change in input voltage
and load. The LQR weight matrices were obtained by trial error method, algebraic
approach and genetic algorithm. To provide better performance with load distur-
bance, LQG controller is proposed. The LQG uses LQR controller with Kalman filter.
Table 1 shows the comparison of transient parameters obtained by different PID
tuning methods.
This comparison shows that LQG-GA method provides the best transient perfor-
mance but if load disturbance of 160% was provided than LQR-GA not able to give
good transient performance. The LQR-GA provides good transient performance till
300% load disturbance. The MATLAB simulink results for different methods are
shown in figures below (Figs. 3, 4, 5, 6, 7, 8, 9 and 10).
5 Conclusion
In this paper, Zeigler–Nichols, LQR and LQG methods of tuning PID controller were
introduced. The LQR and LQG methods provide optimal design of PID controller.
These tuning methods have been successful to provide improved transient parameter
like rise time, settling time, peak overshoot and steady-state error. The designed PID
controller is capable of compensating the effect of change in input voltage and load
disturbances.
Fig. 3 Simulink result of P type Z-N controller a Output Voltage b Inductor Current
Fig. 4 Simulink result of LQR trial and error method a Output Voltage b Inductor Current
6 Future Scope
The optimal control methods LQR and LQG may also be used to improve transient
performance of boost, buck-boost and Cuk converter. These methods will able to
provide optimal solution of problem. The weight matrices of LQR state feedback
system may also be obtained by using neural network and fuzzy controller. Future
research can be done to design robust and discrete PID controller using optimal
theory.
Fig. 5 Simulink result of LQR trial and error method with load disturbance a Output Voltage
b Inductor Current
Fig. 6 Simulink result of LQR algebraic approach a Output Voltage b Inductor Current
Fig. 7 Simulink result of LQR algebraic approach with load disturbance a Output Voltage
b Inductor Current
Fig. 8 Simulink result of LQR TE a Output Voltage b Inductor Current

Fig. 9 Simulink result of LQG-AA a Output Voltage b Inductor Current
Fig. 10 Simulink result of LQG-GA a Output Voltage b Inductor Current
References
1. Mohan N, Undeland TM, Robbins WP (2003) Power electronics converters. John Wiley and
Sons, Inc., Applications and Design
2. Giovanni Beccuti A, Papafotiou G, Morari M (2005) Optimal control of the boost dc-dc
Converter. In: 14th IEEE conference on decision and control, and the European control
conference. CDC-ECC
3. Srivastava S, Kumar Y, Misra A, Thakur SK, Pandit VS (2013) Optimum design ofbuck
converter controller using LQR approach. In: 15th international conference onadvanced
computing technologies, ICACT
4. Tan RH, Hoo LY (2015) DC-DC converter modeling and simulation using state spaceapproach.
IEEE Conf Energy Convers CENCON 2015(2):42–47
5. Camacho-Solorio L, Sariñana-Toledo A (2014) I-LQG control of DC-DC boost converters.
In: 11th International conference on electrical engineering, computing scienceand automatic
control. CCE
6. Li Y, Ang KH, Chong GCY (2006) Pid control system analysis and design. IEEE Control Syst
Mag 26(1): 32–41
7. Li H, Ye X (2010) Sliding-mode PID control of DC-DC converter. In: Proceedings of the 2010
5thIEEE conference on industrial electronics and applications, ICIEA pp 730–734
8. Kumar C, Lal S, Patra N, Halder K, Reza M (2012) Optimal controller design for inverted
pendulum system based on LQR method. IEEE Int Conf Adv Commun Control Comput
Technol ICACCCTno. 978:259–326
9. Poodeh MB, Eshtehardiha S, Kiyoumarsi A, Ataei M (2007) Optimizing LQR and poleplace-
ment to control buck converter by genetic algorithm. In: ICCAS internationalconference on
control, automation and systems, pp 2195–2200
10. Tan GQ, Chen YH, Gu L (2014) LQR based optimal PID control for buck converter. Appl
Mech Mater 687–691:3221–3226
Chapter 33
Comparative Analysis of Different
Maximum Power Point Techniques
for Solar Photovoltaic Systems
Shyam Lal Vishnoi and Kusumlata Agarwal
1 Introduction
The photovoltaic cell has nonlinear I–V characteristics and the direct extraction of
energy is lower from the SPV module. To extract the maximum power from the
SPV module, it is necessary to match the PV source to the load so that the operating
point of PV module coincides with a maximum power point. This job will be done
by DC–DC converter. To extract the maximum power from SPV module, there are
so many different MPPT techniques from which output of the solar panel can be
maximized. According to maximum power transfer theorem that power transferred
will be maximum when the load resistance is equal to source resistance so in all
techniques input resistances are so adjusted that it can be made equal to load resistance
so that maximum power takes place. So to do the source resistance equal to load
resistance duty cycle of the power semiconductor is adjusted by all these MPPT
techniques. By adjusting the duty cycle power transferred can be maximized. So
these MPPT techniques continuously track the peak power point and according to
that, some action is being taken on duty cycle whether it has to increase or decrease.
There are so many MPPT techniques available in literature out of which four different
MPPT listed below will be discussed in this paper.
• Perturb and Observe (P&O) Method
• Incremental Conductance (INC) method
• Fractional open circuit voltage method (FOCV)
• Fractional short circuit current method (FSSC)
S. L. Vishnoi (B) · K. Agarwal

Department of EE, Jodhpur Institute of Engineering and Technology, Jodhpur, Rajasthan, India
e-mail: shyamlal209e@gmail.com
K. Agarwal
e-mail: kusumagarwal@jietjodhpur.ac.in

https://doi.org/10.1007/978-981-15-8820-4_33
358 S. L. Vishnoi and K. Agarwal
Fig. 1 Cuk converter interfacing a resistive load to an SPV panel
These methods are widely used although these methods are having inherent deficien-
cies despite the fact these methods are widely used. The researchers are consistently
trying to improve the existing methods with the help of artificial neural network
methods. This paper presents the comparison of different MPPT techniques with
variable load in MATLAB SIMULINK environment and also hardware validation
of two MPPT is also incorporated [1, 2].
2 DC–DC Cuk Converter
The proposed converter in this dissertation is Cuk converter. All the works have been
carried out on this converter only. This converter has input side capacitance so that
nonlinear output voltage will be nearly constant for DC–DC Cuk converter and a
IRF840 MOSFET is used for switching operation and inductor is an integral part of
this converter [3–5] (Fig. 1).
2.1 Working of Cuk Converter
See Fig. 2.
2.2 Input–Output Relationship

Vo = V P V . D
1−D
& Io = 1−D
D
IPV
33 Comparative Analysis of Different Maximum Power Point Techniques … 359
Fig. 2 Current roots during (a) DT s time and (b) (1-D)T s time
2.3 Terminal Resistance

2
1− D
RT = RL
D
3 Perturb and Observe (P&O) MPPT
The Perturb and Observe technique is very robust, very popular, very simple and
efficient techniques generally used in most of the industries for maximum power
point racking. So in this technique, voltage and current are sensed by voltage and
current sensor respectively and then power is calculated based on the data available
from sensor. So this power is compared with the previous stored data of power and
according to this duty cycle is perturbed. Here the flow chart of P&O method is
shown below for MPPT [6, 7] (Fig. 3).
Fig. 3 Flow chart for perturb and observe (P&O) method
4 Incremental Conductance Methods
Here by referring the I–V characteristics of a PV panel by drawing the It v/s Vt

characteristics and P–V characteristics as shown below. And at a peak power point,
a vertical line is drawn which cut on I–V characteristics at operating point at MPP,
i.e., at Pm, at this point Pm the slope of power dd VP is zero. To the left of operating
point, the slope of power is positive and the right of the operating point the slope of
power is negative [8, 9] (Fig. 4).
So in a nutshell it can be written as.
( ddVItt + VItt ) > 0 Left of MPP.
d It
d Vt
+ VItt = 0 at MPP.
d It
d Vt
+ VItt < 0 Right of MPP.
5 Fractional Open Circuit Voltage MPPT Method
Vm
= k = 0.7
Voc
For Voc to be at 100% then Vm will be approximately 70%. It is varying from 70

to 90% so once a particular PV panel is chosen then with the help of datasheet, the
ratio of Vm to Voc can be obtained which will be a constant [10, 11] (Figs. 5, 6).
Fig. 4 (a) I–V & P–V characteristics of SPV Module, (b) Flow chart for incremental conductance
method
It
a
DC
Vt to
Ro
DC
Isc
Vm
Fig. 5 (a) Typical PV system, (b) I–V & P–V curve for that PV system
iT iT 1
IO Ppv Isc
RT
+
+ Im 1
DC-DC Converter Iscn RTn
VT VO
RO Imn
- -
VT
(a) Voc
(b)
Fig. 6 (a) A typical PV interfacing system, (b) A typical I–V and P–V curve at different insolations
6 Fractional Short Circuit Current MPPT method-
The relationship between Isc and Im for a particular set or insolation so one can
use the following assumption is much stronger than the voltage assumption that was
made in the previous method so it can be written as
Im Im
= = k = 0.9
Isc Isc
7 Designed Parameters Value of Cuk Converter
8 PV System Design and Simulation in MATLAB/Simulink
See Fig. 7.
Fig. 7 Simulink model of PV system with MPPT algorithm

Fig. 8 Simulink block diagram of subsystem FOCV MPPT block
Table 1 Designed Parameters Value of Cuk Converter

Parameter D L1 L2 C1 C2
Theoretical Value 0.6663 8.35mH 16.68mH 177.59µF 2.50 µF
Practical value 0.6663 L 1 = 8.53mH, r 1 = L 2 = 18.766mH, r 2 200 µF 47 µF
0.62Ω = 0.285Ω
8.1 MATLAB Simulation of Fractional Open Circuit Voltage

(FOCV) MPPT Technique
The MATLAB/Simulink of DC–DC Cuk converter with fractional open circuit

voltage (FOCV) MPPT is shown in Fig. (8) below. The FOCV is applied after
0.1 s so that the effect of applied MPPT can be seen [12, 13] (Fig. 8 and Table 1).
8.2 MATLAB Simulation of Fractional Short Circuit Current

(FSSC) MPPT Technique
See Fig. 9.
Fig. 9 Simulink block diagram of subsystem FSSC MPPT block

Fig. 10 Simulink block diagram of subsystem PO based MPPT controller
8.3 MATLAB Simulation of Perturb and Observe (P&O)

MPPT Technique
See Fig. 10.
8.4 MATLAB Simulation of Incremental Conductance (INC)

MPPT Technique
See Fig. 11.
Fig. 11 Simulink block diagram of subsystem INC based MPPT controller

8.5 Performance Parameters of Fractional Open Circuit

Voltage (FOCV) MPPT Technique
All the performance parameters are measured with the two different cases as one
is for constant insolation of 1000 W/m2 for variable load and another for different
insolation levels like as 1000 W/m2 , 800 W/m2 and 600 W/m2 for constant load of
40. FOCV is applied after 0.1 s so that the effect of MPPT can be seen [14, 15, 16]
(Figs. 12 and 13).
Fig. 12 (a) PV panel current (Ipv) and output load current (Io) with FOCV MPPT, (b) PV panel
voltage (Vpv) output load voltage (Vo) with FOCV MPPT
Fig. 13 (a) PV panel power (Ppv) output load power (Po) with FOCV MPPT, (b) Tracking of duty
cycle waveform with FOCV MPPT
8.6 Performance Parameters of Fractional Short Circuit

Current (FSSC) MPPT Technique
Figure 4.18 shows the PV panel current which is fairly constant at 2.82 at insolation
of 1000w/m2 and at 2.3 A at 800w/m2 insolation. The PV panel voltage is fairly
constant at 18 V which is very close to the peak power point voltage i.e. 17.7. The
output voltage is 42 V at 1sun and 38 V at 0.8 sun and 33 V at 0.6 sun [17, 18]
(Figs. 14 and 15).
Fig. 14 (a) PV panel current (Ipv) output load current (Io) with FSSC MPPT, (b) PV panel voltage
(Vpv) output load voltage (Vo) with FSSC MPPT
8.7 Performance Parameters of Perturb and Observe (P&O)

MPPT Technique
See Figs. 16 and 17.

Fig. 15 (a) PV panel power (Ppv) output load power (Po) with FSSC MPPT, (b) Tracking of duty
cycle waveform with FSSC MPPT
Fig. 16 (a) PV panel current (Ipv) and output load current (Io) with P&O MPPT, (b) PV panel
power (Ppv) and output load power (Po) with P&O MPPT
Fig. 17 P–V and I–V tracking curve with P&O MPPT applied
Fig. 18 (a) PV panel current (Ipv) of DC–DC Cuk converter with P&O MPPT for different loads,
(b) PV panel voltage (Vpv) of DC–DC Cuk converter with P&O MPPT for different loads
8.8 Performance Parameters of P&O MPPT with Different

Loads
See Figs. 18, 19 and 20.
8.9 Performance Parameters of INC MPPT with Different

Loads
The figures below show the PV panel current and output current at different loads.
Initially, MPPT has been disabled and then after some interval MPPT is being enabled
so that effect of MPPT can be seen [19, 20] (Fig. 21).
Fig. 19 (a) PV panel power (Ppv) of DC–DC Cuk converter with P&O MPPT for different loads,
(b) Output current (Io) of DC–DC Cuk converter with P&O MPPT for different loads
9 Hardware Implementation of MPPT Techniques
There are so many MPPT techniques that can be implemented using hardware but
in this thesis, two MPPT techniques are being implemented using hardware which
is perturb and observe (P&O) MPPT and incremental conductance (INC) MPPT
technique. For implementing these MPPT techniques, Arduino Uno Atmega320P
[15] will be used as a microcontroller that will give the PWM signal to the DC–DC
Cuk converter [21, 22, 23] (Fig. 22).
MOSFET driver circuit uses the IC TLP250 which is basically a MOSFET driver
IC. Arduino is hardware cum software tool which is used as a controller just as micro-
controller because it uses the microcontroller IC Atmega320P of ATMEL company
and the programming is very easy in the Arduino. For sensing of data, voltage and
current sensors are being used. One can use the ACS712-5A as a current sensors and
Fig. 20 (a) Output voltage (Vo) of DC–DC Cuk converter with P&O MPPT for different loads,
(b) Output power (Po) of DC–DC Cuk converter with P&O MPPT for different loads
ACS712-25 V as voltage sensors [24, 25, 26, 27] (Figs. 23, 24 and 25) (Tables 2, 3,
4 and 5).
10 Comparative Analysis Experimental Results

of Different MPPT Techniques
Table 6 shows the comparative analysis of the results obtained from the MPPT
techniques applied to the DC–DC Cuk converter.
Fig. 21 (a) Input current (Ipv) and output current (Io) of DC–DC Cuk converter with INC MPPT
for different loads, (b) Input power (Ppv) and output power (Po) of DC–DC Cuk converter with
INC MPPT for different loads
The two MPPT techniques have been discussed and implemented on hardware and
its results are obtained. Based on these results, a comparison has to be made between
the results of these two MPPT techniques, i.e., P&O and INC MPPPT techniques
applied to the DC–DC Cuk converter. Table 6 shows the comparative analysis of the
results of P&O and INC MPPT applied to Cuk converter [26, 27, 28].
However, these output parameters are not constant, it is varying near these values.
From the comparative analysis, it is seen that power output available in case of INC
MPPT is more and smooth that can be seen from the graph of output power obtained
[6, 6].
a b
Fig. 22 (a) Duty cycle waveform generated by IC TLP250 on DSO, (b) Top view of hardware
validation of DC–DC Cuk converter
Fig. 23 The tracking of duty with the change in step size of duty with P&O MPPT
Fig. 24 The PV panel power with the change in step size of duty with P&O MPPT
Fig. 25 Performance parameter with INC MPPT for variable load
Table 2 Available insolation and ISC and VOC 40 load resistance
S. No Load R () Voc (Volts) Isc (Amp) Insolation (G) w/m2 Temperature T (°C)
1 40 19.30 V 2.645A 842 w/m2 56 °C
Table 3 Output side parameters for constant load of 40 with P&O MPPT applied
MPPT Enable Load R () Vo (Volts) Io (Amp) Po (w) Duty
Before MPPT 40 12.82 V 0.325A 4.04 W 0.404
After MPPT 40 31.08 V 0.773A 24.02 W 0.70
Variation After MPPT 40 28.80 V–31.19 V 0.72A–0.78A 22 W– 0.66–0.72
24 W
Table 4 Available insolation and ISC and VOC for variable load (INC)
S.No Load R () Voc (Volts) Isc (Amp) Insolation (G) w/m2 Temperature T (ºC)
1 25 19.68 V 2.033A 647 w/m2 56 ºC
2 35 19.70 V 2.197A 700 w/m2 54 ºC
3 45 19.64 V 2.277A 725 w/m2 60 ºC
Table 5 Output side parameters for variable load with INC MPPT applied
MPPT Load R () Vo (Volts) Io (Amp) Po (w) Duty
After MPPT 25 23.90 V 0.958A 22.90 W 0.65
After MPPT 35 28.43 V 0.807A 22.94watt 0.69
After MPPT 45 32.18 V 0.716A 23.04 W 0.72
Table 6 Comparative analysis of experimental results of MPPT techniques

MPPT MPP Insolation Average VOUT IOUT POUT Average
TECHNIQUE MODE (W/m2) Temperature (Volt) (Amp.) (Watts) Duty (D)
(°C)
P&O Disable 850 56 12.82 0.325 4.04 0.404
Enable 850 56 31.08 0.773 24.02 0.70
INC Disable 850 56 12.45 0.309 3.84 0.404
Enable 850 56 32.89 0.827 27.20 0.72
11 Conclusions
In this paper, four different MPPT techniques have been presented using MATLAB
SIMULINK and two MPPT techniques have been experimentally realized. A compar-
ative analysis of all four MPPT techniques has been made The simulation is done
using MATLAB/SIMULINK to find out the performance of MPPT techniques under
different weather conditions, i.e., steady-state and dynamic weather conditions. From
the results obtained, it has been noticed that IC is found to be the best method for
steady-state weather conditions as far as the efficiency and the power losses are
concerned. However, the implementation of IC is complex because it requires two
sensors for proper operation. FSSC and FOCV are the easiest to implement at soft-
ware level but they require additional hardware arrangements for measuring the Isc
and Voc PO is easy to implement in terms of both embedded software and hardware
but is less efficient under both steady and dynamic weather conditions. The compar-
ison presented in this paper will enrich the research community and will help in
future research in this field [25,27–31].
References
1. Priyadarshini, Rai S (2014) Design, modelling and simulation of a PID controller for buck
boost and cuk converter. Int J Sci Res (IJSR) 3(5)
2. Kushwaha BK, Narain A (2012) Controller design for cuk converter using model order
reduction. In: International conference on power, control and embedded systems
3. Han BC, Kim M, Lee S, Lee JS (2015) Dynamic modeling and integral sliding mode controller
design for cuk converter under load variation. In: 9th international conference on power
electronics-ECCE Asia
4. Mokal BP, Vadirajacharya K (2017) Extensive modeling of DC-DC cuk converter operating
in continuous conduction mode. In: International conference on circuits power and computing
technologies [ICCPCT]
5. Rao P, Siraswar V, Pimple BB (2017) Efficient implementation of MPPT and comparison of
converter for variable load Solar PV system. In: 2nd IEEE international conference on electrical,
computer and communication technologies (ICECCT)
6. Safari A, Mekhilef S (2011) Simulation and hardware implementation of incremental conduc-
tance mppt with direct control method using cuk converter. IEEE Trans Industr Electron
58(4):1154–1161
7. Panigrahi BK, Thakura PR (2017) Implementation OF cuk converter with MPPT. In: 3rd
international conference on advances in electrical, electronics,information, communication
and bio-informatics (AEEICB17)
8. Samantara S, Roy B, Choudhury RSS, Jena B (2017) Modeling and simulation of integrated
CUK converter for grid connected PV system with EPP MPPT hybridization. In: IEEE power,
communication and information technology conference (PCITC) Siksha ‘O’ Anusandhan
University, Bhubaneswar, India
9. Sahu TP, Dixit TV (2014) Modelling and analysis of perturb and observe and incremental
conductance MPPT algorithm for pv array using cuk converter. In: IEEE student’s conference
on electrical, electronics and computer science
10. Kumar R, Choudhary A, Koundal G, Singh A, Yadav A (2017) Modelling/simulation of MPPT
techniques for Photovoltaic systems using matlab. Int J Adv Res Comput Sci Softw Eng 7(4)
11. Hadeed Ahmed Sher, Ali Faisal Murtaza, Abdullah Noman, Khaled E. Addoweesh, Kamal
Al-Haddad, “A New Sensorless Hybrid MPPT Algorithm Based on Fractional Short-Circuit
Current Measurement and P&O MPPT”, IEEE transactions on sustainable energy, 2015
12. Enslin JHR, Wolf MS, Snyman DB, Swiegers W (1997) Integrated photovoltaic maximum
power point tracking converter. IEEE Trans Industr Electron 44(6):769–773
13. Elgendy MA, Atkinson DJ, Zahawi B (2016) Experimental investigation of the incremental
conductance maximum power point tracking algorithm at high perturbation rates. IET Renew
Power Gener 10(2):133–139
14. Tan RHG, Teow MYW (2014) A comprehensive modeling, simulation and computational
implementation of buck converter using MATLAB/Simulink. In: IEEE conference on energy
conversion (CENCON), pp 37–42
15. Makni W, Ben Hadj N, Samet H, Neji R (2016) Design simulation and realization of solar battery
charge controller using Arduino Uno. In: 17th IEEE international conference on sciences and
techniques of automatic control and computer engineering (STA), pp 635-639
16. Chy DK, Khaliluzzaman M, Karim R (2017) Analysing efficiency of DC-DC converters joined
to PV system run by intelligent controller. In: IEEE international conference on electrical,
computer and communication engineering (ECCE), pp 457–462
17. Halder T (2011) Charge controller of solar photo-voltaic panel fed (SPV) battery. In: India
international conference on power electronics 2010 (IICPE2010)
18. Kollimalla SK, Mishra MK (2014) Variable perturbation size adaptive P&O MPPT algorithm
for sudden changes in irradiance. IEEE Trans Sustain Energy 5(3):718–728
19. Singh B, Kumar R (2016) Solar PV array fed brushless DC motor driven water pump. In: IEEE
6th international conference on power systems (ICPS)
20. Popa D-L, Nicolae M-S, Nicolae P-M, Popescu M (2016) Design and simulation of a 10
MW photovoltaic power plant using MATLAB and Simulink. In: IEEE international power
electronics and motion control conference (PEMC), pp 378-383
21. Sathya P, Natarajan R (2013) Design and implementation of 12V/24V closed loop boost
converter for solar powered led lighting system. Int J Eng Technol (IJET) 5
22. Sahin ME, Okumus HI (2013) Small signal analyses and hardware implementation of a
buck-boost converter for renewable energy applications. In: IEEE international conference
on renewable energy research and applications, pp 330-335
23. Geethanjali MN, Sidram MH (2017) Performance evaluation and hardware implementation of
MPPT based photovoltaic system using DC-DC converters. In: IEEE international conference
on technological advancements in power and energy (TAP Energy)
24. Lin B-T, Lee Y-S (1997) Power-factor correction using Cuk converters in discontinuous-
capacitor-voltage mode operation. IEEE Trans Industr Electron 44(5):648–653
25. Bandyopadhyay A, Parui S (2017) Dynamical behaviour of Cuk converter fed from a
photovoltaic source. In: IEEE Calcutta Conference (CALCON), pp 397-402
26. Thiago A. Pereira , Thamires P. Horn , Walbermark M. dos Santos, Samir A. Mussa , Denizar
C. Martins and Roberto F. Coelho, “Electrical characterizer of photovoltaic modules using
the DC/DC Ćuk converter”, IEEE International Conference on Industrial Technology (ICIT),
pp. 954 - 959, February 2018
27. Singh SP, Gautam AK, Tripathi SP, Kumar B (2017) Performance comparison of MPPT
techniques using Cuk converter for photovoltaic energy conversion system. In: IEEE 3rd
international conference on computational intelligence and communication technology (CICT)
28. Sun Z, Yang Z (2017) Improved maximum power point tracking algorithm with Cuk converter
for PV systems. J Eng 2017(13):1676–1681
29. Lina B-R, Huangb C-L, Tsaoa F-P (2009) Integrated Cuk-forward converter for photovoltaic-
based LED lighting. Int J Electron 96(9):943–959
30. Ferrites and Accessories, [Online], Available: https://www.tdk-electronics.tdk.com/download/
519704/069c210d0363d7b4682d9ff22c2ba503/ferrites-and-accessories-db-130501.pdf
31. Datasheet of TLP-250 Manufactured by TOSHIBA, [Online], Available: https://toshiba.sem
icon-storage.com/info/docget.jsp?did=16821
Chapter 34
MATLAB/Simulink-Based Tracking
of Maximum Power Point
in a Generalized Photovoltaic Module
by Using DC-DC Boost Converter
Yogesh Joshi and Vinit Mehta
1 Introduction
Today, the whole world is facing an enormous drawback of electric power shortage,
which is of great concern because of the conventional methods of generating electrical
energy. In India, per capita electrical energy consumption is increasing fast during
recent few years; it was around 347.5 kWh in 1992, which has reached to 1075 kWh
in the year 2016 and around 1122 kWh in the year 2017 [1]. In general, from the
year 2017, the contribution and shares of various power sectors to meet our energy
demand and financial growth of India are; coal 59%, Hydro 14%, fossil fuels 17%
including 8% of gas, and nuclear 2% [2]. The high dependency of coal, oil, and fossil
fuels for energy generation results in greater energy insecurities in future. With the
rapidly growing requirements of energy and as per the energy forecast, the electrical
energy requirement for the year 2021–2022 will be around 1750 kWh and around
2300 kWh for the 2026–2027 year [3]. Therefore to meet out this demand, energy
system needs to be more efficient and sustainable with minimum environmental and
ecological impacts for harvesting more energy by using Renewable Energy Sources.
Among all the Renewable Energy Sources, energy generation from photovoltaic
effect is considered as more effective and sustainable resource, as it uses sun as a
natural energy resource, which is eco-friendly, clean, inexhaustible, abundance in
nature, free and energy supplied by it consistently. A major initiative has been taken
up by the Government of India to promote and increase maximum utilization of solar
energy by proposing “National Solar Mission” in the year 2008, which was retitled
Y. Joshi (B)
Electrical Engineering Department, J.I.E.T, Jodhpur, Rajasthan, India
e-mail: yogi.jo89@gmail.com
V. Mehta
Electrical Engineering Department, J.N.V.U, Jodhpur, Rajasthan, India
e-mail: vinit741@gmail.com

https://doi.org/10.1007/978-981-15-8820-4_34
382 Y. Joshi and V. Mehta
Table 1 Phases and targets of (JNNSM)

Application Phase-1 year Phase-2 year Phase-3 year
(2010–2013) (2013–2017) (2017–2022)
Rooftop stand-alone solar 0.2 GW 1 GW 40 GW
plants
Grid connected solar plants 1–2 GW 4–10 GW 60 GW
Total 100 GW
as “Jawaharlal Nehru National Solar Mission” (JNNSM) in the year 2010. The main
aim of this operation is to deploy 20 Giga Watt (GW) of solar energy by the end of
the year 2022. Three-phase approach is required for the execution of this operation.
Different phases and goals of JNNSM are mentioned in Table 1 [4].
The energy obtained by the sun based upon the photovoltaic framework depends
upon solar irradiance, cell temperature, and the voltage formed in the solar photo-
voltaic cells. Besides the fact that the solar photovoltaic system gives a new direction
in the energy generation process, it also has some major drawbacks, which are.
1. Higher initial cost;
2. Low power conversion efficiency;
3. Electric power produced by sunlight-based photovoltaic framework is highly
affected by the atmospheric conditions; variation in atmospheric conditions
results in continuous change in electric power.
Figure 1 provides more clarity about the whole concept and working of how to
separate most extreme power from the photovoltaic module using the MPPT system
and DC-DC power converter.
As the climatic conditions for the whole duration of the day are not constant and
certain parameters like solar irradiance and temperature are continuously varying
all through the day, under such conditions it is difficult to find peak point where
maximum power occurs. MPPT is an impedance matching and time-varying tech-
nique that interfaces changing source and possibly the fluctuating load. Among
several MPPT algorithms, two more effective, fast and widely used MPPT methods
Fig. 1 Schematic diagram of maximum power tracking using DC-DC converter

34 MATLAB/Simulink-Based Tracking of Maximum Power Point … 383
that can work accurately and precisely in every single imaginable condition in order
to get maximum power are.
1. Perturb and Observe Method (P&O Method),
2. Incremental Conductance Method (INC Method).
Perturb and Observe (P&O) and Incremental Conductance (INC) method are the
two most optimizing and fast-tracking MPPT algorithms to track maximum power
point by repeatedly updating the working voltage of the photovoltaic system. MPPT
and PWM (Pulse-Width Modulation) controller helps in adjusting the duty cycle
of the power DC-DC power converters with a fixed step size to transfer maximum
power from the source to load.
Results obtained after simulation with conclusions are discussed in the later
stage. Solar photovoltaic module, MPPT controller, and DC-DC power converter are
designed and simulated using MATLAB/Simulink and simulated graphs such as I-V
and P–V characteristics, graph of output power from DC-DC converter are obtained.
Results of a simulated solar photovoltaic module are compared and validated with
the actual results of the photovoltaic module.
2 State of the Art
Solar photovoltaic panels comprise of a number of solar photovoltaic modules which

when connected with suitably designed power conditioning units such as Maximum
Power Point Tracking (MPPT) controller and DC-DC converters produces maximum
output power. Previous studies suggest that various investigations have been made
on solar photovoltaic systems for different applications to gain as much benefits as
possible from solar energy.
Some of the research works and papers presented on solar photovoltaics, simula-
tion, and modeling of photovoltaic module in MATLAB/Simulink, interconnection
of electrical load and photovoltaic module with a DC-DC power converter and imple-
mentation of Maximum Power Point Tracking (MPPT) strategies and methods are
briefly conferred in this section.
Shyam. B and P. Kanakasabapathy had given an insight of various government
policies, acts, plans, and strategies such as “Jawaharlal Lal Nehru Solar Mission”
(JNNSM), JNNSM phases and targets initiated by Indian government agencies for
country’s economic growth and development. The approach toward development in
the area of Smart Grid in India is additionally explained in their paper [1]. Marcelo
Gradella Villalva, Jonas Rafael Gazoli, and Ernesto Ruppert Filho, had determined
the three basic parameters of the photovoltaic model which are open-circuit voltage,
short-circuit current, and maximum power point from the I-V characteristic curves.
A photovoltaic array is designed and simulated in MATLAB/Simulink and using the
simulated results mainly the output power of the photovoltaic array is matched and
validated with the practically available photovoltaic array [5].
Huan-Liang Tsai, Ci-Siang Tu, and Yi-Jie Su had given an insight into how
maximum power can be achieved through simulating and analyzing the proposed
model by using power electronic devices [8]. Weidong Xiao, William. G. Dunford,
and Antoine Capel had given a data-based approach for photovoltaic (PV) modeling.
Nonlinear mathematical equations were solved for getting I-V characteristics of
photovoltaic cells [9].
J. A. Cow and C. D. Manning had proposed a specific circuit-based simu-
lated photovoltaic array model for the development of advanced solar photovoltaic
(PV) conversion system [10]. S. Sheik Mohammed and D. Devaraj had presented
a research study that provides a more comprehensive and detailed description
on developing generalized photovoltaic module by means of simulation platform
in MATLAB/Simulink and validating the simulated module with the monetarily
accessible MSX 60 photovoltaic module [11].
Pandiarajan. N and Ranganath Muthu had portrayed about building up an exact
power electronic circuit model necessary for designing and simulating photovoltaic
integrated system [13]. Mei Shan Ngan, Chee Wei Tan have discussed two categories
of MPPT algorithms, one is direct and another is indirect method [16].
Ting-Chung Yu and Yu-Cheng Lin had analyzed three most extensively
used tracking methods specifically Perturb & Observe (P&O) algorithm, Incre-
mental Conductance (INC) algorithm and Hill climbing algorithm. They described
the complete development of photovoltaic system utilizing MPPT methods in
MATLAB/Simulink by combining solar module and DC-DC converter with
maximum power point tracking algorithms [17]. Trishan Esram and Patrick L.
Chapman had given a comparative analysis of maximum power point tracking strate-
gies on the photovoltaic array. They had discussed different methods for tracking
maximum power point condition and their implementation [18].
In general, many simplifying assumptions have been made in the various proposed
methods for maximum power point tracking. The aim of this paper is to develop a
comprehensive computational methodology for building MATLAB/Simulink model
to simulate with ‘Perturb & Observe’ and ‘Incremental Conductance’ algorithm
incorporating DC-DC boost converter for getting maximum output power.
The output power of the solar photovoltaic system is highly influenced by rapidly
varying climatic conditions such as solar radiation and temperature. Solar photo-
voltaic system exhibits nonlinear I-V and P–V characteristics under different climatic
conditions; operating point on P–V characteristics curve fluctuates continuously from
its maximum position. The process of operating a photovoltaic array or module at
or closer to maximum power point condition is known as Maximum Power Point
Tracking (MPPT).
The method, which is to be applied for tracking maximum power point should
be easy to implement, take lesser time to track, it should have adequate convergence
speed, and price of implementation ought to be less. Various techniques have been
proposed for tracking the Maximum Power Point (MPP), among these techniques,
Perturb and Observe (P&O) and Incremental Conductance (INC) methods are most
widely utilized techniques and are applied for tracking maximum power point.
(a) Perturb and Observe (P&O) Method:
P&O method is the most broadly utilized methodology for tracking maximum power
due to its easy approach and very few parameters are required for the implementa-
tion of this method. The power generated by the photovoltaic module or array is
perturbed, observed, and analyzed through iteration methods to locate maximum
power point condition. Solar photovoltaic module or array voltage is perturbed peri-
odically with increment and decrement in terminal voltage and comparing the output
power received from photovoltaic with the previous perturbed cycle. The basis of this
control algorithm is the slope (dP/dV) on the P–V characteristic curve as depicted in
Fig. 2 [17].
Condition I: When the working voltage of photovoltaic module or array is
perturbed in a particular direction and dP/dV > 0 (i.e., left aspect of the curve), then
slope dP/dV is positive and direction of perturbation is toward Maximum Power
Point (MPP) and P&O algorithm continues to perturb photovoltaic module or array
voltage in the same direction.
Condition II: When dP/dV < 0, (i.e., right aspect of the curve), then slope dP/dV
is negative and operating point moves far away from Maximum Power Point (MPP)
and during this condition P&O algorithm reverses the direction of perturbation.
Condition III: At dP/dV = 0, condition of maximum power point is reached.
Though maximum power can be obtained at dP/dV = 0, still some power is dissipated
during tracking as the operating point does not remain stable and it oscillates around
maximum power point. The basic operating procedure of P&O technique is given in
Fig. 2.
Fig. 2 Schematic diagram of P–V characteristic curve showing different positions of dP/dV
Advantages of P&O method:

1. Simple in structure and easy to implement.
2. Low computational demand.
3. Implementation cost is less.
4. Only one sensor, i.e., voltage sensor is required to sense the photovoltaic module
or array voltage.
Disadvantages of P&O method:

1. Rapidly varying climatic conditions has a negative effect on the P&O method as
algorithm starts tracking Maximum Power Point (MPP) in the wrong direction.
The reason behind this is when atmospheric conditions changes, then simulta-
neously solar irradiance also changes rapidly due to which Maximum Power
Point (MPP) deviates slightly from its position in either direction; P&O algo-
rithm concludes this change as a change due to perturbation and accordingly in
the successive iterations it starts perturbing in the wrong direction.
2. P&O method cannot find out exactly that when it has reached maximum power
point (MPP) and operating point continuously oscillates around Maximum Power
Point (MPP), which leads to some energy loss in the process.
3. Slow response speed at the event of changes in solar irradiance.
To overcome the limitations of the P&O algorithm, a new technique for tracking
MPP has been established known as “Incremental Conductance methods”..
(b) Incremental Conductance (INC) Method:
The main concept of Incremental Conductance Method depends upon the way that
slope of power curve at Maximum Power Point (MPP) is zero in the photovoltaic
module; it is positive on left of Maximum Power Point (MPP) and negative on the
right of the MPP, this can be explained as [18]
⎧ dP
⎪
⎪ = 0, at MPP
⎪
⎪ dV
⎪
⎨
dP
> 0, left of MPP (1)
⎪ dV
⎪
⎪
⎪
⎪
⎩ dP < 0, right of MPP
dV
The above equation can be stated in terms of voltage and current as [16–18].
dP d(IV)
=
dV dV
dI
=I+V (2)
dV
When the operating point is at MPP then dP/dV = 0; Eq. (2) can be rearranged as
dI
I+V =0
dV
Therefore,
I + V dV
dI
=0
dI I
=− (3)
dV V
From the above Eq. (3) [18], it can be concluded that left side of the equation
shows photovoltaic module’s incremental conductance (dI/dV) and right side of the
equation shows photovoltaic module’s instantaneous conductance (I/V). Therefore,
during the tracking process, operating point continue to perturb until incremental
conductance value (dI/dV) equals the instantaneous conductance value or (I/V). The
other two conditions when the slope of photovoltaic module power curve is on the
left of the MPP (dP/dV > 0) and when it is on the right of the MPP (dP/dV < 0) can
be expressed in terms of voltage and current by rearranging Eq. (2) as
dI I
>− (4)
dV V
dI I
<− (5)
dV V
Here, dI and dV represent change in current and voltage before and after the
increment respectively, whereas I and V represent instantaneous values of voltage
and current, respectively. The above two Eqs. (4) and (5) helps in determining the
direction of perturbation so that operating point moves towards the direction of
maximum power point. All these equations can be summarized as [18].
⎧ dI I dP
⎪
⎪ = − ; when = 0; at MPP
⎪
⎪ dV
⎪
⎨
V dV
dI I dP
> − ; when > 0; at left of MPP (6)
⎪ dV
⎪ V dV
⎪
⎪
⎪
⎩ dI < − I ; when dP < 0; at right of MPP
dV V dV
Figure 3 shows how the operating point in the Incremental Conductance Method
perturbs to reach Maximum Power Point (MPP). The pattern and approach of pertur-
bation in the incremental conductance method is the same as that of perturb and
observe method. From Fig. 3, it can be observed that when the operating point is in
the left side of MPP, the direction of perturbation is positive whereas the direction of
perturbation has to be changed (or reversed) when the operating point is at the right
of MPP.
Condition I: When (dI = 0 and dV = 0) that means there is no change in voltage
and current; it concludes that operating point is on the maximum position and MPPT
Fig. 3 Basic concept of

incremental conductance
method on P–V curve
still operates on maximum power point. Atmospheric conditions under this condition
remains unchanged.
Condition II: When only change in current is accounted and there is no change
in voltage; (dV = 0 and dI changes). Change in dI indicates a change in atmospheric
conditions results in a change in solar irradiance, which leads to a change in maximum
power point. When dI > 0 then there is a rise in the amount of solar irradiance which
results in voltage of maximum power point.
Condition III: When current and voltage both changes and both do not equal to
zero (dI/dV = 0). In case, if both voltage and current are changing then Eqs. (4)
and (5) helps in determining the direction of perturbation to reach maximum power
point. If dI/dV > -I/V, then dP/dV > 0, this indicates that operating point on P–V
curve is toward the left of the MPP.
Figure 4 explains the whole process, about how MPPT in combination with DC-
DC converters is used to extract maximum power from the solar photovoltaic module
or array.
Figure 4 shows that voltage Vpv and current Ipv obtained from solar panel are
input to control algorithm. Control algorithm may be P&O method or Incremental
Conductance method, implemented to track maximum power point condition by
perturbing duty cycle. PWM (Pulse-Width Modulation) controller helps in adjusting
and monitoring duty cycle (α) used for triggering purpose of power electronic switch
Fig. 4 Interconnection of
solar photovoltaic module
and load through DC-DC
converter using control
mechanism
Fig. 5 Schematic diagram of Buck converter and waveforms
(i.e., MOSFET) of DC-DC Converter. Duty cycle (α) of the converter is controlled
in such a way that the source will send maximum power to the load.
(c) Types of DC-DC Converters
1. Buck Converter,
2. Boost Converter, and
3. Buck–Boost Converter.
c1. Buck Converter
Buck Converter is generally used when photovoltaic module or array voltage (input
voltage) is high and battery voltage (output voltage) is low. It is also termed as
step-down converter (Fig. 5).
In Buck Converter, switch S operates at a higher frequency to produce a chopped
output voltage. By adjusting on/off duty cycle of the switching, power flow is
controlled. The average output voltage is given by [19]
Vo ton
= =D (7)
Vi T
c2. Boost Converter
Boost Converter is generally used when a photovoltaic module or array voltage

(input voltage) is low and battery voltage (output voltage) is high. It is also known
as a step-up converter (Fig. 6).
The average output voltage for boost converter is given as [20]
Vo T 1
= = (8)
Vi Toff 1−D
c3. Buck–Boost Converter

Fig. 6 Schematic diagram of boost converter and waveforms
Fig. 7 Schematic diagram of Buck–Boost converter and waveforms
Buck–Boost Converter generally used when battery voltage (output voltage) is

either high or low than the input voltage. It can be used as a step-up or step-down
converter (Fig. 7).
The average output voltage for Buck–Boost converter is given by [21]
Vo ton D
= = (9)
Vi toff 1−D
The MATLAB/Simulink Models of Photovoltaic Module and Implementation of

MPPT Methods are as follows:
Fig. 8 MATLAB/Simulink Blocks of Perturb and Observe (P&O) MPPT Method
Fig. 9 MATLAB/Simulink Blocks of Incremental Conductance (INC) MPPT Method with DC-DC
boost converter
ELDORA-250P, 60 cell polycrystalline solar photovoltaic module manufactured

by Vikram Solar Pvt. Ltd., is proposed as a reference model for evaluating and vali-
dating the simulated photovoltaic model designed in MATLAB/Simulink. ELDORA-
250P is a highly efficient solar photovoltaic module comprising 60 polycrystalline
solar cells, ideally used in rooftop and ground-mounted applications.
Simulated results are compared with the practically and commercially available
ELDORA-250P module for validating the proposed simulated module. An ideal
ELDORA-250P solar photovoltaic module is developed in MATLAB/Simulink envi-
ronment by considering series resistance Rs = 0 and shunt resistance Rsh = ∞.
I-V Characteristics of the proposed model ELDORA-250P at different solar irradi-
ance and at constant temperature of 25 °C are shown in Fig. 10, which are used as
a reference for validating the characteristics to be obtained from the photovoltaic
module designed in this thesis.
Fig. 10 V-I characteristic

curves of ELDORA-250P at
constant temperature 25 °C
and varying solar irradiance
According to the specification of ELDORA-250P solar photovoltaic module, basic

parameters, which are required to design a photovoltaic module at Standard Test
Condition (STC), are given in Table 2 of the manufacturer’s datasheet.
The output power of a solar photovoltaic module is affected mainly due to change
in atmospheric conditions and the two major environmental factors, responsible for
influencing output power of solar photovoltaic module, are power density of sunlight,
Table 2 Commercial photovoltaic module ELDORA-250P datasheet at STC with solar irradiance
1000 W/m2 and temperature 25 °C
Module parameters ELDORA-250P
Maximum power, Pm (W) 250
Maximum voltage, Vm (V) 30.58
Open circuit voltage, Voc (V) 37.55
Short circuit current, Isc (A) 8.71
NOCT 45 °C ± 85 °C
i.e., solar irradiance (W/m2 ) and working temperature (°C). Simulated Model of
ELDORA-250P is tested for two different conditions using MATLAB/Simulink,
which are.
1. Constant temperature and varying solar irradiance,
2. Constant solar irradiance and varying module temperature.
Parameters obtained after simulation at constant solar irradiance (1000 W/m2 )
and varying module temperature (25 °C, 35 °C, 45 °C, 55 °C, 65 °C, and 85 °C) are
given in Table 3.
Simulated results of photovoltaic module parameters obtained from simulated I-
V and P–V characteristic graphs are compared against corresponding actual results
of commercially available ELDORA-250P module parameters which are given in
Table 4.
Table 3 Module parameters extracted after simulation at different temperature level and at constant
solar irradiance of 1000 W/m2
Module parameters Temperature (°C)
25 °C 35 °C 45 °C 55 °C 65 °C 85 °C
Maximum power, Pm (W) 257.6 246.6 238.3 228.1 216.8 198.1
Maximum voltage, Vm 32.1 29.38 28.88 28.26 27.55 24.38
(V)
Open-circuit voltage, Voc 37.2 36.04 34.88 33.72 32.55 30.2
(V)
Short-circuit current, Isc 8.71 8.74 8.78 8.81 8.84 8.91
(A)
Table 4 ELDORA-250P commercial module validation with the simulated module at STC with
solar irradiance of 1000 W/m2 and at 25 °C temperature
Module parameters ELDORA-250P ELDORA-250P % Error = (Actual
(Actual Results) (Simulated Results) Results-Simulated
Results)/Actual
Results × 100%
Maximum power, Pm 250 257.6 3.04%
(W)
Maximum voltage, Vm 30.58 32.1 4.97%
(V)
Open-circuit voltage, 37.55 37.2 0.93%
Voc (V)
Short-circuit current, Isc 8.71 8.71 0.00%
(A)
NOCT 45 °C ± 85 °C
Table 5 Comparison of simulated results of P&O and INC MPPT methods

MPPT method Output Power Output voltage Output current
Pout (Watt) Vout (Volts) Iout (Ampere)
(P&O) method 797.47 199.68 3.99
INC method 897.75 211.86 4.23
From Table 4, it can be easily understood that simulated results obtained at STC
are approximately nearer to the actual results with small deviations, which may be
considered as negligible and are articulated in the form of % error.
DC-DC boost converter is modeled in MATLAB/Simulink for enhancing power
to be received from ELDORA-250P simulated photovoltaic module using Perturb &
Observe (P&O) and Incremental Conductance (INC) MPPT controller; both MPPT
controllers are also designed in MATLAB/Simulink controlling the switching action
of MOSFET. Both the MPPT controllers control duty cycle used for triggering
MOSFET of the DC-DC boost converter. Load resistance of 50 is connected
across Simulink model of the DC-DC boost converter.
DC-DC boost converter designed in MATLAB/Simulink consists of two capaci-
tors, one with a rating of 450 μF and another with a rating of 330 μF, one inductor
of 120 μH, one resistive load of 50 , one power diode, and one MOSFET. Both the
MPPT methods are tested using simulated ELDORA-250P photovoltaic module at
standard test condition (STC) having solar irradiance of 1000 W/m2 and temperature
of 25 °C.
Results obtained after simulation are as follows:
Simulated results achieved from both the MPPT methods after simulation are
analyzed in tabulated form in Table 5.
5 Conclusion
The proposed photovoltaic module built in MATLAB/Simulink illustrates and veri-

fies nonlinear I-V and P–V characteristic curves of the photovoltaic module. Essential
parameters are obtained from simulated I-V and P–V graphs which are mentioned in
Table 2. After analyzing the simulated I-V and P–V characteristics, it is concluded
that with the increase in solar irradiance, both open-circuit voltage Voc and short-
circuit current Isc of photovoltaic module increases, thereby increasing the output
power of the photovoltaic module. The reason behind the increase in output power is
the logarithmic dependence of open-circuit voltage Voc on solar irradiance whereas
short-circuit current Isc is directly proportional to the solar irradiance.
When photovoltaic module is built in MATLAB/Simulink under different temper-
ature levels and at constant solar irradiance and simulated I-V and P–V character-
istic graphs are generated. Essential parameters are obtained from simulated I-V
and P–V graphs which are mentioned in Table 3. After analyzing the simulated
I-V and P–V characteristics, it is concluded that with the increase in module’s
working temperature, short-circuit current Isc of photovoltaic module increases
marginally whereas open-circuit voltage Voc of photovoltaic module decreases dras-
tically, thereby decreasing the output power of the photovoltaic module. Hence, it is
concluded that temperature has a significant effect in determining solar photovoltaic
module’s efficiency and performance of the photovoltaic module is highly affected
by the increase in the module’s working temperature.
Advantage of using DC-DC boost converter amongst the other converters is that in
case if MPPT circuit fails to operate and switching action of MOSFET will not take
place, therefore in this condition load is directly connected to the panel and panel
will still supply power to the load but at less efficiency. Other DC-DC converters
cannot fulfil this condition. This paper describes only about implementation of boost
converters.
References
1. Shyam B, Kanakasabapathy P, Renewable energy utilization in india-policies, opportunities

and challenges. In: 2017 IEEE international conference on technological advancements in
power and energy (TAP Energy).
2. Central Electricity Authority of India, “Draft-National Electricity Plan (2016),[Online], https://
www.cea.nic.in/reports/committee/nep/nep_dec.pdf. [Accessed: July 6, 2017].
3. Central Electricity Authority of India, National Electricity Plan (2018). https://www.cea.nic.
in/reports/committee/nep/nep_jan_2018.
4. Jawaharlal Nehru National Solar Mission Document
5. Villalva MG, Gazoli JR, Filho ER (2009) Comprehensive approach to modeling and simulation
of photovoltaic arrays. IEEE Trans Power Electron 24(5)
6. Lynn PA (2010) Electricity from sunlight: an introduction to photovoltaics. Wiley, New York,
p 238
7. Markvart T (1994) Solar electricity. Wiley, New York
8. Tsai H-L, Tu C-S, Su Y-J (2008) Member, IAENG, Development of generalized photovoltaic
model using Matlab/Simulink. In: WCECS 2008. San Francisco, USA
9. Xiao W, Dunford WG, Capel A (2004) A novel modeling method for photovoltaic cells. In:
35th annual IEEE power electronics specialists conference. Aachen, Germany
10. Cow JA, Manning CD (1996) Development of model for photovoltaic arrays suitable for use
in simulation studies of solar energy conversion systems. In: Power electronics and variable
speed drives, Conference Publication No. 429, ©IEE
11. Sheik Mohammed S, Devaraj D (2014) Simulation and analysis of stand-alone photovoltaic
system with boost converter using MATLAB/Simulink. In: International conference on circuit,
power and computing technologies [ICCPCT]
12. Sheik Mohammed S (2011) Modeling and simulation of photovoltaic module using
MATLAB/Simulink. Int J Chem Environ Eng 2(5)
13. Pandiarajan N, Muthu R (2011) Development of power electronic circuit oriented model of
photovoltaic module. Int J Adv Eng Technol IJAET II(IV):118–127
14. Pandiarajan N, Ramaprabha R, Muthu R, Application of circuit model for photovoltaic energy
conversion system. (Manuscript Title—Revised), Research Article, https://www.hindawi.com/
journals/ijp/2012/410401.
15. Sridhar R, Jeevananathan, Selvan NT, Banerjee S (2010) Modeling of PV array and performance
enhancement by MPPT algorithm. Int J Comput Appl (0975–8887) 7(5)
16. Ngan MS, Tan CW (2011) A study of maximum power point tracking algorithms for stand-alone
photovoltaic systems. IEEE Appl Power Electron Colloquium (IAPEC)
17. Yu T-C, Lin Y-C, A study on maximum power point tracking algorithms for photovoltaic
systems
18. Esram T, Chapman PL (2007) Comparison of photovoltaic array maximum power point tracking
techniques. IEEE Trans Energy Convers 22(2)
19. https://en.wikipedia.org/wiki/Buck_converter (Access on Date 2 June 2019, Time 07:45 p.m.)
20. https://en.wikipedia.org/wiki/Boost_converter (Access on Date 2 June 2019, Time 07:45 p.m.)
21. https://en.wikipedia.org/wiki/Buck–boost_converter (Access on Date 2 June 2019, Time 07:45
p.m.)
Chapter 35
Design and Simulation
of MPPT-Operated DC-DC Flyback
Converter Used for Solar Photovoltaic
System
Ranjana Choudhary and Shrawan Ram Patel
1 Introduction
One of the major problems faced by the whole world is related to the environment
like pollution, global warming, waste disposal, greenhouse gases, CO2 emissions and
many but the reason behind these problems are basic needs of humans. Causes of co2
emission are natural as well as human sources. Natural sources like decomposition,
respiration, ocean release, and the human sources are like production of cement,
deforestation, burning of fossil fuels. The demand for electricity has increased the use
of fossil fuel like coal, lignite, oil, gas. To fulfil the requirement of electricity, power
plants are necessary. In India, almost 60% of the electricity is generated from thermal
power plants which use coal as a fuel. There are some other power plants also in
which gas and oil are used to burn and generate electricity but it causes Co2 emission
and resultantly, harming our environment. So it is necessary to switch to some other
sources to generate electricity which can be environment friendly. The need for
human to generate electricity using environment-friendly sources has brought this
attention toward natural sources like sun, water, air. Electricity generation using these
sources requires a technique which should be the most efficient. For bulk generation
as well as for small-scale generation or domestic-level generation, solar energy is
the most suitable form of energy. Being particular about solar energy, there are
many techniques to extract solar energy but the demand is for the most efficient and
economic technique which is the key objective of the paper. In literature, it was found
that the Maximum Power Point Tracking Method is the most efficient and economic
R. Choudhary (B)
EE Department, JIET-Jodhpur, Mogra Khurd, Rajasthan, India
e-mail: ranjana.choudhary2017@gmail.com
S. R. Patel
Department of Electrical Engineering, JIET-Jodhpur, Mogra Khurd, Rajasthan, India
e-mail: shrawanram.patel@jietjodhpur.ac.in

https://doi.org/10.1007/978-981-15-8820-4_35
398 R. Choudhary and S. R. Patel
to harness the solar energy using PV Panel. PV modules, MPP Techniques along
with net metering can be used at consumer level so that the surplus electrical energy
can be fed to the AC Grid. This prototype will require DC-DC converter, DC-AC
converter, etc. DC-DC conversion with solar PV power has become very noteworthy
for efficient and economical energy production to exploit the maximum solar power
possible at every moment, increasing profits and sustainability. The objective of this
paper is “design and simulation of MPPT guided an isolated dc-DC Flyback converter
used for solar PV system”.
There are several methods for exploiting PV solar energy and many existing
conventional techniques had incorporated the use of DC-DC converter along with
DC-AC converter (inverter) for AC voltage applications. This type of converter
topologies has two foremost features as reported in the literature. The first one is
where the transformerless inverter is used which has several advantages like abridged
size and low cost with high working efficiency. In this topology, the main drawback is
that voltage boosting is not achieved as per the need of AC Voltage application or grid
applications of voltage range 85−265 V AC. There is another problem of isolation
between low voltage circuit (PV side) and high voltage circuit (AC side) and also
the problem of grounding in the solar cell side is serious in case of transformerless
topologies. The second feature is one where the transformer or coupled inductor
system is used and the operating frequency is high. This feature will mitigate the
problem of grounding of PV Panel and paved a way of boosting the voltage. Also,
the problem of earth parasitic capacitance which will become a source of leakage
current is avoided by using an isolation transformer. Since the cost of a line frequency
transformer is very high and due to its large volume and weight, the high frequency
(say 100 kHz) transformer is preferred [1, 2]. In line with this dissertation work,
several authors have worked on Flyback converter with different aspects as follows.
Xiong et al. [3] have implemented the improved maximum or peak power point
tracking in a PV system based on Flyback converter. A versatile simulation using
MATLAB-based model is constructed for the output voltage behaviours or charac-
teristics of a particular PV system varies nonlinearly with different solar insolation
(G W/m2 ), panel temperature.
Daniel Calheiros de Lemos [4] has presented MPPT-guided DC-DC Flyback
Converter, which fed the solar energy into the grid. The converter used is situated
within the family of DC-DC boost/buck power electronic converters. The algorithm
presented here is based on Perturb and Observe technique, where it only controls the
power generated by the panel, but the algorithm presented controls the power and the
current of the panel. It has been found from the literature survey that several assump-
tions have been made in order to simulate and to realize the performance of DC-DC
Flyback converter using different MPPT techniques in the various proposed methods.
Many authors have tried to deal with the DC-DC Flyback converters interfacing a
load with either Solar PV panel/module/array or rectified DC supply in order to
extract the maximum power from the given source. The goal of this paper is to build
the software simulation of Flyback converter, being operated by InC MPPT tech-
nique, and obtain the performance of converter with varying insolation, temperature
and load impedance.
35 Design and Simulation of MPPT-Operated DC-DC Flyback Converter … 399
Patil and Kumar [5] have been presented the analysis of open- and closed-loop
control of DC-DC Flyback converter. The main objective is to get the voltage mode
control in the closed-loop model using a robust PID controller by which the working
efficiency is augmented by 9%. The designed models of the open- and closed-loop
DC/DC Flyback converter are verified using a MATLAB simulation.
Prabhakaran and Prasad [6] have designed and analysed an AC-DC and DC-
AC isolated Flyback converter without the practice of bridge circuit. The proposed
Flyback inverter is driven in continuous conduction (where inductor current is contin-
uous) Modes with low apex current and superior efficiencies. The comparative inves-
tigation of the proposed circuit/topology with conventional circuit consists of AC-DC
and DC-AC converter is addressed in the simulation of the proposed converter and
also implemented using MATLAB Simulink.
2 Basics of Flyback Converter
The Flyback converter is derived from the buck–boost converter with the coupling
inductor. The advantage of Flyback converter is that it can provide voltage multi-
plication by varying the number of turns of the coupled inductor and can provide
multiple output DC/DC converters at power levels of 150 W or less. The design
parameters of a Flyback converter is discussed in this chapter.
3 Working of Flyback Converter
The Flyback converter is similar to buck–boost converter in design and working. The
only difference in construction is of coupling inductor. During ON state, when the
MOSFET switch is Turn-ON or closed, the primary winding of the coupled inductor
is connected directly to the PV module or source. The current flowing in the primary
winding of coupled inductor increases and magnetic flux in the coupled inductor also
increases. In this duration, energy is stored in the primary winding of the coupled
inductor and the voltage which was induced in the secondary winding of the inductor
is inverted. Hence, the diode is in the blocked stage or reversed-biased. During ON
state, the capacitor connected at the output side delivers electrical energy to the load.
During OFF state, when the MOSFET switch is Turn-OFF, the current of primary
and magnetic flux decreases. The voltage of secondary inductor is positive and the
diode comes into conduction state which allows the current to flow to the load circuit.
The energy from the secondary coupled inductor is supplied to the capacitor and then
supplied to the load and this cycle is continued. The operation of Flyback converter
can be continuous conduction mode or discontinuous mode but in this paper, the
continuous conduction mode of Flyback converter is chosen so that maximum energy
can be harnessed from solar PV module/array (Fig. 1).
3.1 Continuous Conduction Mode of Flyback Converter
A DC-DC converter that provides isolation between the input side and output side
is shown in Fig. 2a. In Fig. 2b, it uses the transformer model which includes the
magnetizing inductance Lm. To understand the operation of the circuit, a simplified
transformer model is considered.
The following additional assumptions for the analysis are made:
• The output capacitor Co is very large and it is providing a constant output voltage
Vo.
• The circuit is in steady-state operation which means that all the voltages and
currents are cyclic in nature, i.e. they are starting and ending at the same points
over one cycle/switching period.
Fig. 1 Circuit diagram of a DC-DC flyback converter
Fig. 2 a Flyback converter. b Equivalent circuit using a transformer model that includes the
magnetizing inductance
Fig. 3 Flyback Converter circuit during a T on b T off
• The duty cycle of the switch is D, ON-time period is of DT and OFF-time period
is of (1-D)T.
• The semiconductor switch and power diode are ideal, giving a zero conduction
loss. The basic operation of the Flyback converter is the same as that of the
buck–boost converter.
• When the switch is closed, energy is stored in Lm. When the switch is open, it is
transferred to the load. This circuit is analysed for both closed and open switch
positions to determine the relationship between input and output.
Various waveforms of Flyback converter are shown in Sect. 3.13. The following
relationships are required to design the Flyback converter (Fig. 3):
• Output Voltage (Vo )

D N2
Vo = V pv (1)
1− D N1
• Inductor Current

Vo 1 N2
ILm = (2)
RL 1− D N1
• Minimum Value of Inductance (Lm )

2
(1 − D)2 R N1
L m_min = (3)
2f N2
• Value of Output Capacitor
D
Co = (4)
R L f. V
Vo
o
4 Parameter Calculation for Flyback Converter
The following steps are required to determine the parameters of the Flyback
converter:
Step 1: With no losses in the Flyback converter, the output voltage across the load
resistance of 40 will be
Vo2
Pin = Pout ⇒ Vmpp Impp =
RL
Vo2
50 = ⇒ Vo = 44.7213 volt
40
If the turn ratio is N1 /N2 = ½, then the duty of Flyback converter will be

D N2 D 2
Vo = Vin ⇒ 44.7213 = 17.7 × ×
1− D N1 1− D 1
D = 0.5581 or 55.81%
Step 2: The average inductor current through the inductor Lm is given by
Vo2 44.72132
I Lm = = = 5.06155A
V pv D R L 17.7 × 0.5581 × 40
Step 3: The minimum value of inductance is

2 2
(1 − D)2 N1 (1 − 0.5581)2 1
L M_ min = R = 40 × = 19.52 m H
2f N2 2 × 50000 2
Step 4: The output capacitance C o is given by
D 0.5581
Co = = = 27.05μF
R L f. V
Vo
o 40 × 50000 × 0.01
5 Subsystem Configuration of Flyback Converter
Using a Flyback converter, the output voltage can be done step-up or down depending
upon the duty ratio of the converter. In this thesis, the proposed Flyback converter
is used as an interface between the load with a PV panel of 50 Wp. The controlling
parameter of the Flyback converter for a fixed value of load resistance and operating
frequency are provided to ‘Pulse Generator’ block to produce the pulse signal. This
pulse signal is given to IGBT gate terminal for its ON/OFF operation. Figures 4
Fig. 4 Simulation circuit
Fig. 5 Maximum power

point tracking block
and 5 show the simulation diagram of Flyback converter interfacing the load to SPV
panel with MPP technique. The PV module is subjected to variable insolation levels
(200 W/m2 to 1000 W/m2 ) and panel temperature is kept constant to observe the
effect of insolation on the performance of the PV module.
6 Experimental Results
The experimental setup for the proposed Flyback DC-DC converter is simulated in
MATLAB, built and tested for effectiveness under varying load impedance, panel
temperature and solar insolation.
6.1 Performance of Flyback Converter at Different Insolation

Levels
The operating voltage of the PV panel and terminal current is varying with insolation
levels. The panel current is increasing with insolation and decreasing with a decrease
in insolation as shown in Table 1.
The input and output power of the Flyback converter with varying insolations is
depicted in Fig. 6.
The efficiency of the Flyback converter with insolation is shown in Fig. 6. At
standard condition (G = 1000 W/m2 and Panel Temperature = 25 °C), the efficiency
of converter is around 93%. From the efficiency curve, it can be stated that the
conversion efficiency of the DC/DC Flyback converter is increasing with increasing
solar insolation and temperature remains constant. But in real world, the temperature
of the PV panel is also increasing with insolation and this effect is highlighted in the
next session.
Table 1 Photovoltaic panel current with different insolations at constant temperature

Insolation (G)W/m2 Current (Ipv ) A
200 0.5527
400 1.127
600 1.687
800 2.276
1000 2.849
Fig. 6 Input and output power of Flyback converter with varying solar insolation
6.2 Performance of Flyback Converter at Different

Temperature Levels
When the PV panel is subjected to both different insolations and temperatures, the
output power is slightly lowered compared to the previous one (when the temperature
is constant) as shown in Table 2.
The compared power of Flyback converter is shown in Fig. 7. From Fig. 7, one can
say that as insolation is increasing the temperature of the PV module is increasing and
this will cause of alleviation of the corresponding output power of the PV module.
And also, with increased insolation, the photogenerated current is increased. With
increased panel current the switching losses and the conduction losses are increased.
Table 2 Output power of flyback converter with different insolation levels and different temperature
levels
Insolation (G) (W/m2 ) Panel temperature (o C) Output power
(Po ) Watts
200 25 8.441
400 30 17.43
600 40 26.23
800 50 34.16
1000 60 41.26
Fig. 7 Output power for

different temperatures and
different insolations
6.3 Performance of Flyback Converter with Varying Load

Conditions
The load resistance, in this paper, was varied from 20 to 50 and the output power
is slightly decreasing as shown in Fig. 8.
The MPPT technique is trying to extract the maximum power from the PV module
for any value of load resistance. Here, the load resistance is increasing and to match
this increased load resistance with characteristics impedance of solar PV panel, the
duty should be increased. If the switching device is Turn-ON for a lonng time then PV
panel can’t be able to feed the maximum available power to the load. The comparative
analysis of the Flyback converter with varying insolation is depicted in Table 3.
(a) (b)
Fig. 8 a Input power and b output power of Flyback converter for varying loads
Table 3 Performance of flyback converter with variable loads at standard conditions

RL Vpv Ipv Ppv Vo Io Po η Duty
20 17.4057 2.8527 49.5854 30.5296 1.5265 47.0289 94.84% 0.4669
25 17.4348 2.8522 49.6668 34.0519 1.3621 46.8063 94.24% 0.4933
30 17.5034 2.8443 49.7361 37.3515 1.2451 46.8081 94.11% 0.5142
35 17.4766 2.8498 49.7566 40.216 1.1490 46.5322 93.52% 0.5326
40 17.5276 2.8405 49.7322 42.8496 1.0712 46.239 92.98% 0.5473
45 17.4783 2.8483 49.7269 45.3022 1.0067 45.9609 92.43% 0.5615
50 17.5384 2.8376 49.7046 47.5855 0.9517 45.6603 91.86% 0.5726
6.4 Performance of Flyback Converter Under Dynamic

Weather Conditions
The output power of the Flyback converter is also increasing by increasing the solar
insolation. The output power is a combined effect of temperature and solar inso-
lation because the photogenerated current is directly proportional to the insolation
but terminal voltage falls logarithmically with an increase in panel temperature.
The power of Flyback converter is obtained for different insolations at different
temperatures as shown in Fig. 9.
The comparative efficiency of Flyback converter with varying insolation and
temperature is depicted in Fig. 9b. The efficiency of the Flyback converter is high
for lower temperature for the same insolation level. The efficiency of the Flyback
converter with variable load is depicted in Fig. 10.
Fig. 9 a Output Power with varying insolations at constant temperature. b Output Power with
varying insolations at varying temperatures
Fig. 10 Efficiency and

power of Flyback converter
versus duty ratio
7 Conclusions
The proposed Flyback DC-DC converter is realized using MATLAB for different
temperature levels, insolation levels and for varying insolation levels. Following
main points are the conclusion for this paper.
(1) With increasing temperature level of solar PV panel, the efficiency of the system
falls down.
(2) With increasing insolation, the efficiency of DC-DC converter is increasing
when the temperature remains constant but in practice it is not possible. There-
fore, the efficiency of the converter will slightly fall with increasing insolation
level.
(3) With increasing load impedance, the duty of converter is increased and this
results in large conduction loss in switch and hence efficiency will decrease.
References
1. Zhao S, Li Q, Lee FC, Li B (2018) High frequency transformer design for modular power
conversion from medium voltage AC to 400V DC. IEEE Trans Power Electron 33(9):7545–7557
2. Shen J-M, Jou H-L, Jinn-Chang Wu (2012) Novel transformer less grid-connected power
converter with negative grounding for photovoltaic generation system. IEEE Trans Power
Electron 27(4):1818–1829
3. Xiong X, Shen A (2016) Improved maximum power point tracking in PV system based on
flyback converter. In: 2015 Chinese automation congress (CAC)
4. Daniel Calheiros de Lemos,MPPT Algorithm for DC-DC Converter for Photovoltaic Panel
Applications, [online available] https://fenix.tecnico.ulisboa.pt/ArtigoDaniel_Lemos_N77168.
pdf
5. Patil B, Kumar P (2016) Performance analysis of a flyback converter. Int J Adv Res Electr
Electron Instrum Eng 5(9)
6. Prabhakaran J, Prasad MB (2017) Design of novel bridgeless AC-DC-AC fly back convertor
using Matlab. Int J Res Appl Sci Eng Technol (IJRASET) 5(III)
Chapter 36
To Improve Power Transfer Capacity
Using TCSC FACTS Controller
Kishore Singh Gehlot and Shrawan Ram
1 Introduction
Electrical networks are interconnected to different generating stations and load

centers according to the existing plan. But load demands on the system are not
constant. With the increase of industrial growth and domestic load, more power is
consumed by the different loads. To fulfill the load demand, either electrical system
network has to be re-evaluated or the power carrying capability of the transmission
line has to be increased. From economic point of view, modification or alteration
of the electric network is costly. Thus the aim is to increase the power carrying
capability of the transmission line.
How to Improve Power Flow:
The various parameters involved in power flow are.
(a) Load angle,
(b) Transmission line impedance, and
(c) Operating variables such as voltage and current.
To control the power flow from one bus to another bus, either of three parameters
to be controlled. Power systems of today are mechanically controlled. Mechanical
switching action is slow and power flow control is not fast enough according to the
load variation. Another problem of mechanical control is that it cannot be initiated
frequently as it leads to wear and tear.
To maintain both dynamic and steady-state operation, the new technology, i.e.,
FACTS (Flexible AC Transmission System) is used which is a power–electronics-
based system. Its main role is to enhance controllability and power transfer capability
K. S. Gehlot (B) · S. Ram

Department of EE, Jodhpur Institute of Engineering and Technology, Jodhpur, Rajasthan, India
e-mail: kishoresinghgehlot7@gmail.com

https://doi.org/10.1007/978-981-15-8820-4_36
410 K. S. Gehlot and S. Ram
in AC systems. FACTS technology uses switching power electronics to control power

flow in the range of tens to hundreds of megawatts.
The various FACTS controller are capable of controlling the interrelated line
parameters and other operating variables as mentioned in this paragraph. Thus
FACTS technology governs the operation of the transmission line by providing shunt
impedance, series impedance, line current, phase angle, voltage, and damping of
oscillations at various frequencies below the rated frequency. By providing added
flexibility, FACTS controllers can enable a transmission line to carry power:
• By maintaining proper insulation of transmission line without over-voltage,
• Up to its thermal rating,
• By maintaining stability in the system,
Thus, the use of FACTS technology increases the power carrying capability of
the existing transmission network which is more economical.
Various FACTS Controllers:
In general, FACTS controllers may be divided into four categories:
• Shunt controller,
• Series controller,
• Combined series-shunt controllers, and
• Combined series-series controllers.
In this paper, it is considered about one of the Series Controller that is Thyristor-
Controlled Series Capacitor (TCSC) to control the power flow in the transmission
line. TCSC is one of the FACTS devices, which consist of a series capacitor bank
shunted by a Thyristor-Controlled Reactor (TCR) in order to provide a smoothly vari-
able series capacitive reactance by varying the firing angle of Thyristor-Controlled
Reactor (TCR).
2 Power Flow Solution of Large Electrical Network
The main objective of a power flow study is to do determination the steady-state

operating condition of the electrical power network. The steady state may be deter-
mined by finding out the flow of active and reactive power throughout the power
system and the voltage magnitude and phase angles at all nodes of the network.
The planning and daily operation of current power systems leads to many power
flow studies. Such information is used to carry out security evaluation analysis, where
the nodal voltage magnitudes and active and reactive power flows in transmission
lines and transformers are carefully observed to assess whether or not they are within
the given operating limits. If the power flow study indicates that there are voltage
magnitudes outside limits at certain points in the network, then appropriate control
actions become necessary to regulate the voltage magnitude. Similarly, if the study
presumes that the power flow in a given transmission line is out of the power carrying
capacity of the line then control action will be taken.
36 To Improve Power Transfer Capacity Using TCSC FACTS Controller 411
Generalized Power Flow Solution For 5-Bus Network: A 5-bus network is

given in the next page in Fig. to analyze the power flow solution and to determine
the active power flow and reactive power flow from each bus. Also, nodal voltage
magnitude and nodal phase angle is determined where these quantities are unknown.
For this reason, a computer program is used to solve the power flow solution.
Data for 5-bus Network
Transmission Line Data
Transmission line From To Resistance in p.u Reactance in p.u Susceptance in p.u

Tline-1 Giral Merta 0.02 0.06 0.06
Tline-2 Giral Sangariya 0.08 0.24 0.05
Tline-3 Merta Sangariya 0.06 0.18 0.04
Tline-4 Merta Piparcity 0.06 0.18 0.04
Tline-5 Merta Jodhpur 0.04 0.12 0.03
Tline-6 Sangariya Piparcity 0.01 0.03 0.02
Tline-7 Piparcity Jodhpur 0.08 0.24 0.05
Generator Bus Data
Bus no Bus type Nodal Nodal phase Active Reactive Generators Generators
voltage in angle in p.u power power reactive reactive
p.u injected in injected in power upper power lower
p.u p.u limits in p.u limit in p.u
Giral-1 Slack Bus 1.06 0 Unknown Unknown 5 −5
Merta-2 Generator 1.00 To be 0.4 To be 3 −3
PV Bus calculated calculated
Load Bus Data
Bus no Bus type Nodal voltage in Nodal phase angle Active power Reactive power
p.u in p.u drawn in p.u drawn in p.u
Merta-2 LoadPQ Bus To be calculated To be calculated 0.20 0.10
Sangariya-3 LoadPQ Bus To be calculated To be calculated 0.45 0.15
Piparcity-4 LoadPQ Bus To be calculated To be calculated 0.40 0.05
Jodhpur-5 LoadPQ Bus To be calculated To be calculated 0.60 0.1
General Parameters: Maximum Iteration = 100, Tolerance = 1e – 12.

Computer Program for Power Flow Solution Using Newton–raphson Method:
For the 5-bus network shown in Fig. 4, a computer program is written using
MATLAB code to solve the power flow equation (Figs. 1, 2, 3 and 4).
Fig. 1 A 5-Bus network for analyzing the power flow solution
Fig. 2 Active and reactive power flow shown at each bus for 5-bus network
Fig. 3 TCSC connected between to bases k and m
Fig. 4 Active and reactive power flow shown at each bus for 6-bus network where one TCSC is
used to control the power flow
Summary of Outputs:
ACTIVE POWER SENT FROM EACH BUS:

GIRAL GIRAL MERTA MERTA MERTA SANGA PCITY
MERTA SANGA SANGA PCITY JODHPUR PCITY JODHPUR
===== ===== ===== ===== ===== ===== =====
89.4458 42.0023 24.5614 27.8124 54.7540 19.3916 6.6320
ACTIVE POWER RECEIVED AT EACH BUS:
MERTA SANGA SANGA PCITY JODHPUR PCITY JODHPR
===== ===== ===== ==== ======= ======= ========
-86.8455 -40.2730 -24.1132 -27.2521 -53.4448 -19.3461 -6.5552
ACTIVE POWER LOSS ON TRANSMISSION LINE:
===== ===== ===== ===== ===== ===== =====
2.6002 1.7293 0.4482 0.5603 1.3092 0.0455 0.0768
REACTIVE POWER SENT FROM EACH BUS:
===== ===== ===== ===== ===== ===== =====
74.1880 17.0217 -2.4910 -1.6897 5.6034 2.8679 0.5491
REACTIVE POWER RECEIVED AT EACH BUS:
MERTA SANGA SANGA PCITY JODHPUR PCITY JODHPR
===== ===== ===== ==== ======= ======= ========
-72.9084 -17.5125 -0.3523 -0.8306 -4.8292 -4.6878 -5.1708
REACTIVE POWER LOSS ON TRANSMISSION LINE:

===== ===== ===== ===== ===== ===== =====
1.2797 -0.4908 -2.8433 -2.5202 0.7742 -1.8199 -4.6216
NET ACTIVE POWER INJECTED/OUTAGE AT GENERATOR/LOAD BUS:
GIRAL MERTA MERTA SANGA PCITY JODHPUR
131.4481 40.2823 -20.0000 -44.9946 -39.9663 -60.0000
NET REACTIVE POWER INJECTED/OUTAGE AT GENERATOR/LOAD BUS:
GIRAL MERTA MERTA SANGA PCITY JODHPUR
91.2097 -61.4856 -10.0000 -14.9969 -4.9692 -10.0000
TOTALGENERATION_ACTIVEPOWER = 171.7304
TOTALGENERATION_REACTIVEPOWER = 29.7241
TOTAL_ACTIVE_LOAD = -165
TOTAL_REACTIVE_LOAD = -40
TOTAL_ACTIVEPOWER_LOSS = 6.7304
TOTAL_REACTIVEPOWER_LOSS = -10.2759
TOTALITERATION = 6
Conclusion:
The summary of the result is superimposed on the given 5-bus network shown in
Fig-6.5 so that at each bus we can find out how much power flows from each bus
and the losses on the transmission line are also mentioned. The summary of output
shows that at each bus, power mismatch equation is satisfied. Also, the power flow
solution converges to a prescribed tolerance of, i.e., 12 within six iterations.
3 Power Flow Control Using Tcsc
As Thyristor-Controlled Series Capacitor (TCSC) will control the power flow in the
transmission line of a large electrical network, here we will model the Thyristor-
Controlled Series Capacitor (TCSC) as a variable reactance which varies in terms of
firing angle of the thyristor.
Power Flow Solution for 6-Bus Network:
Power flow solution for the network where Thyristor-Controlled Series Capacitor
(TCSC) is used in the network is solved in a similar way using Newton–Raphson (NR)
method. In the next, a 6-Bus network is drawn, in which a TCSC is used in between
Sangriya to Pokran. By writing a MATLAB code, we study how the Thyristor-
Controlled Series Capacitor is effective for controlling the specified amount of active
power in between two buses.
Data for 6-bus Network in Which Tcsc-1 is Connected:
Transmission Line Data for 7 Lines
Transmission line Fm To Resistance in p.u Reactance in p.u Susceptance in p.u

Tline-6 Pokran Piparcity 0.01 0.03 0.02
Tline-7 Piparcity Jodhpur 0.08 0.24 0.05
Generator Bus Data
Bus no Bus type Nodal Nodal Active Reactive Generators Generators

voltage phase power power reactive reactive
in p.u angle in injected injected in power power
p.u in p.u p.u upper lower limit
limits in in p.u
p.u
Giral-1 Slack bus 1.06 0 unknown unknown 5 −5
PV bus calculated calculated
Load Bus Data
Bus no Bus type Nodal voltage Nodal phase Active power Reactive power
in p.u angle in p.u drawn in p.u drawn in p.u
Merta-2 LoadPQ bus To be calculate To be 0.20 0.10
calculate
Sangariya-3 LoadPQ bus To be calculate To be 0.45 0.15
calculate
Piparcity-4 LoadPQ bus To be calculate To be 0.40 0.05
calculate
Jodhpur-5 LoadPQ bus To be calculate To be 0.60 0.1
calculate
TCSC’s DATA
Total no of TCSC = 1.
Connected between: Sangariya-3 to Pokran-6.
Capacitive reactance of TCSC = 9.375e-3.
Inductive reactance of TCSC = 1.625e-3.
Initial Firing angle = 145 degree.
Firing angle lower limit = 90 degree.
Firing angle upper limit = 180 degree.
Active power to be controlled = 0.21 p.u
General Parameters:
Maximum Iteration = 100,
Tolerance = 1e – 12.
Computer Program for Power Flow Control Using Tcsc (thyristor- Controlled
Series Capacitor):
To solve the power flow equation of the above network, which contains one TCSC
to control the required amount of active power between POKRAN and PIPARCITY,
a program is written in MATLAB code.
Summary of Output:

GIRAL GIRAL MERTA MERTA MERTA POKRAN PCITY
===== ===== ===== ===== ===== ===== =====
88.7927 42.6632 25.5906 26.7005 54.1990 21.0060 7.1711
===== ===== ===== ===== ===== ===== =====
-86.2091 -40.8924 -25.1076 -26.1804 -52.9147 -20.9535 -7.0853
===== ===== ===== ===== ===== ===== =====
2.5836 1.7708 0.4830 0.5201 1.2843 0.0524 0.0858
===== ===== ===== ===== ===== ===== ===== 74.3796
16.9521 -2.6660 -1.5331 5.6513 2.5142 0.4444
REACTIVE POWER RECEIVED AT EACH BUS::::
===== ===== ===== ===== ===== ===== =====
-73.1461 -17.3287 -0.0832 -1.0971 -4.9491 -4.3160 -5.0509
===== ===== ===== ===== ===== ===== =====
1.2335 -0.3766 -2.7493 -2.6302 0.7021 -1.8018 -4.6065
GIRAL MERTA MERTA SANAG PCITY JODHPUR POKRAN
===== ===== ===== ===== ===== ===== =====
131.4559 40.2810 -20.0000 -66.0000 -39.9629 -60.0000 21.0060
GIRAL MERTA MERTA SANAG PCITY JODHPUR POKRAN
===== ===== ===== ===== ===== ===== ===== 91.3317 -
61.6940 -10.0000 -17.4119 -4.9687 -10.0000 2.5142
TOTALGENERATION_ACTIVEPOWER = 171.7369
TOTALGENERATION_REACTIVEPOWER = 29.6377
ACTIVE_POWER_INJECTED_IN_TCSC = 21.0000
REACTIVE_POWER_INJECTED_IN_TCSC =2.4119
FINAL_FIRINGANGLE = 148.4675
TOTAL_REACTANCE = -0.0216
TOTAL_ITERATION = 8
Result-1:
Figure 7.5 is modified and reproduced in Fig. 7.3, in which one TCSC is connected
between Sangariya and Pokran to control 21 MW of power flow from Sangariya to
Piparcity. The power flow solution is obtained in 8 iterations to a power mismatch
tolerance of 1e-12. The power flow results are shown in figure.
Since the TCSC cannot generate active power, there is an increase in active power
flow from GIRAL bus to SANGARIYA bus (i.e., from 42.00 MW to 42.66 MW).
At the same time, there is an increase of active power flow from MERTA bus to
SANGARIYA bus (i.e., from 24.56 MW to 25.59 MW). In total, there is an increase
of active power flow from SANGARIYA to PIPARCITY (i.e., from 19.39 MW to
21 MW).
It should be remarked that transmission line from SANGARIYA to PIPARCITY
is series compensated by the use of TCSC-1 and there is an increase of active power
flow from 19.38 MW to 21 MW, which is just under 8% active power increase.
Thus TCSC with firing angle control provides a good series compensation in the
transmission line for controlling the active.
Power Flow Solution of 7-Bus Network:
The network shown in Fig consists of two TCSC for controlling the power flow.
One TCSC is connected in between Sangariya and Pokran. Other is connected in
between Piparcity and Boranada. The MATLAB code which is written for one TCSC
is now modified with additional data for TCSC-2 and executed. The power flows vary
in the transmission line and it is different from the case used for one TCSC.
Data for 7-bus Network in Which Tcsc-1 and Tcsc-2 is Connected:
Transmission Line Data
Transmission line Fm To Resistance in p.u Reactance in p.u Susceptance in p.u
Tline-6 Pokran Piparcity 0.01 0.03 0.02
Tline-7 Boranada Jodhpur 0.08 0.24 0.05
Generator Bus Data

Bus no Bus type Nodal Nodal phase Active Reactive Generators Generators
voltage in angle in p.u power power reactive reactive
p.u injected in injected in power upper power lower
p.u p.u limits in p.u limit in p.u
Giral-1 Slack bus 1.06 0 unknown Unknown 5 −5
PV bus calculate calculated
Load Bus Data
Bus no Bus type Nodal voltage Nodal phase Active power Reactive power
in p.u angle in p.u drawn in p.u drawn in p.u
Merta-2 LoadPQ bus To be To be 0.20 0.10
calculated calculated
Sangariya-3 LoadPQ bus To be To be 0.45 0.15
Piparcity-4 LoadPQ bus To be To be 0.40 0.05
Jodhpur-5 LoadPQ bus To be To be 0.60 0.1
TCSC’s DATA
Total no of TCSC = 2.
Data for TCSC-1
Connected between: From Sangariya-3 to Pokran-6.
Capacitive reactance of TCSC -1 = 9.375e-3.
Inductive reactance of TCSC-1 = 1.625e-3.
Data for TCSC-2
Connected between: From Piparcity-4 to Boranada-7.
Capacitive reactance of TCSC -2 = 9.375e-3.
Inductive reactance of TCSC-2 = 1.625e-3.
General Parameters:
Maximum Iteration = 100,
Tolerance = 1e – 12.
Computer Program for Power Flow Control Using Tcsc (thyristor- Controlled
Series Capacitor):
To solve the power flow equation of above network which contains one TCSC to
control the required amount of active power between POKRAN and PIPARCITY, a
program is written in MATLAB code.
Summary of Outputs:
SUMMARY OF OUTPUTS:
GIRAL GIRAL MERTA MERTA MERTA POKRAN BNADA
MERTA SANGA SANGA PCITY JODHPUR PCITY JODHPUR ===== =====
===== ===== ===== ===== =====
89.2434 42.7349 25.4970 43.1498 38.3180 21.0033 22.9577
===== ==== ======= ===== ========
-86.6484 -40.9829 -25.0142 -41.8647 -37.6175 -20.9519 -22.3825
===== ===== ===== ===== =====
2.5951 1.7520 0.4828 1.2851 0.7005 0.0514 0.5752
===== ===== ===== ===== =====
74.2474 16.0179 -3.8013 -7.0318 10.0700 -0.1455 -4.4705
REACTIVE POWER RECEIVED AT EACH BUS:
===== ==== ======= ===== ========
-72.9821 -16.4735 1.0360 6.4911 -11.0271 -1.6681 1.0271
===== ===== ===== ===== =====
1.2653 -0.4556 -2.7653 -0.5408 -0.9571 -1.8136 -3.4434
GIRAL MERTA MERTA SANGA PCITY JODHPUR POKRAN BNADA
===== ===== ===== ===== ==== ======= ====== ====
131.9784 40.3164 -20.0000 -65.9971 -62.8167 -60.0000 21.0033 22.9577
GIRAL MERTA MERTA SANGA PCITY JODHPUR POKRAN BNADA
===== ===== ===== ===== ==== ======= ====== ====
90.2652 -63.7452 -10.0000 -15.4376 4.8229 -10.0000 -0.1455 -4.4705
TOTALGENERATION_ACTIVE POWER = 172.2948
TOTALGENERATION_REACTIVE POWER = 31.3430
Result-2
From Fig. 7.5, we observe that TCSC-1 provides series compensation in the transmis-
sion line between SANGARIYA to PIPARCITY bus. This is because active power
flow from SANGARIYA to PIPARCITY is increased from 19.39 MW (shown in
Fig. 6.5) to 21 MW. At the same time the active power flow from GIRAL to
SANGARIYA is increased from 42.00 MW to 42.73 MW and from MERTA to
Fig. 5 Active and reactive power flow shown at each bus for 7-bus network where two TCSC is
used to control the power flow
SANGARIYA is increased from 24.56 MW to 25.49 MW as TCSC doesn’t generate

any active power (Fig. 5).
Similarly from Fig. 7.5, we observe that TCSC-2 provides series compensation
in the transmission line between PIPARCITY to JODHPUR. This is because active
power flow from PIPARCITY to JODHPUR is increased from 6.63 MW to 22 MW. At
the same time, the active power flow from MERTA to PIPARCITY is increased from
27.81 MW to 27.81 MW and MERTA to JODHPUR is decreased from 54.75 MW
to 38.31 MW as TCSC does not generate any active power.
Also the power mismatch equation is satisfied at each bus, after using two TCSC.
Thus from the analysis, it is very clear that both TCSC provides effective series
compensation in the two different transmission lines and specified amount of active
power is controlled.
Conclusion:
From the Result-1 and Result-2, we reached a conclusion that how effectively
Thyristor-Controlled Series capacitor (TCSC) can control the active power flow
between buses. With the use of TCSC, a specified amount of power can be transferred
from one bus to other as TCSC does not consume or generate active power.
4 Conclusion and Further Scope of Work
Summary of Work
The objective of the whole work is to control the power flow in the power system.
This can be achieved by knowing the various parameters which are concerned in the
power flow in the transmission line I. For each parameter, we have discussed how
we can control the active and reactive power flow in the transmission line. For each
parameter, we have applied in MATLAB application software to analyze the power
flow.
As Thyristor-Controlled Series Capacitor (TCSC), is a series compensator is used
in the transmission line to control the active power flow, so we discussed the principle
of series compensation and simulation block is used to verify the truthiness, how a
capacitor is an effective element to decrease the series reactance.
In further discussion, we have analyzed the characteristics of TCSC. By using
MATLAB application, we came to know how effectively TCSC can control the
current and active power flow in the transmission line by varying the firing angle of
TCSC.
A practical electrical network is having a large number of buses. Thus in this
work, a 5-bus imaginary network is considered for finding the power flow solution.
Newton–Raphson method is used to solve this network. By using MATLAB code
for this network, power flow between each bus is determined. Power flow solution
of this network satisfies the power mismatch equation at each bus.
Jacobian matrix is determined for 6-bus network in which one TCSC is used in
between two buses. Again power flow explanation is determined for this network
and we came to know that it also satisfies the power mismatch equations at each bus.
At the same time, a specified amount of active power is allowed to flow between two
buses by using the TCSC. In the next section again, two TCSC is used to analyze
the power flow in different transmission lines. Here also MATLAB code is used for
finding the power flow solution of the given electrical network.
Conclusion
From the execution of MATLAB code, we reached at a conclusion that Thyristor-
Controlled Series Capacitor is one of the fast acting power electronic controllers
which can provide current and power flow control in the transmission line by varying
its firing angle. Thus Thyristor-Controlled Series Capacitor (TCSC) can be used as a
series capacitor to decrease the overall transmission line reactance. Depending on the
development of power transfer preferred at that time, without affecting other system-
performance criteria, series compensation can be varied by Thyristor-Controlled
Series Capacitor (TCSC). Thus, Thyristor-Controlled Series Capacitor (TCSC) is
one of the important FACTS controller, which increases the overall power transfer
capacity in the transmission line.
Further Scope of Work

Works on this topic never end with unlimited applications. It can be applied to
damping of the power swings from local and interconnected area oscillations, voltage
regulation of local network, reduction of short-circuit current, etc. Various research
works are going on control interaction between multiple TCSC. Also, SVC-TCSC
can be combined and used within power systems to enhance inter-area stability. In this
work MATLAB code is used for 5-bus network for power flow solution. MATLAB
code can also be used for designing a physical 20-bus network to determine the exact
power flow from one bus to another. Though in this discussion Mi Power software
is used for 5-bus systems, it is capable of being used in a large bus network. Further
study can be done on modern power system of TCSC on Fault Component Distance
Protection and Impact of TCSC on the Protection of Transmission Lines. Thus TCSC
can be used in many fields of the power system.
References
1. Saddat H (2002) Power system analysis (MATLAB concept). Tata McGraw-Hill, Edition
2. Rashid MH (2004) Power electronics (concept of Flexible AC Transmission Systems), 3rd edn.
Pearson Education
3. Narain G (2001) Hingorani/Laszlo Gyugyi, Understanding FACTS, 1st edn. IEEE press
4. Nagrath/DP Kothari IJ (2003) Power system engineering (concept of series and shunt
compensation). Tata McGraw-Hill
5. C.L. Wadhwa, “Electrical Power Systems (compensation in power system)”, New Age
International, 3rd Edition,2004
6. P.Moore and P. Ashmole, “Flexible ac transmission systems : part 4-advanced FACTS
controller,” Power Engineering Journal, April 1998,pp.95–100
7. E Acha/VG Agelidis/T J E Miller, “Power Electronic Control in Electrical Systems”, Newnes
Power Engineering Series,Ist Indian Edition,2006
8. Fuerte-Esquivel CR, Acha E, Ambriz-Perez H (Feb.) A thyristor controlled series compen-
sator model for the power flow solution of practical power networks. Power Systems, IEEE
Transactions on Power Dlivery 15(1):58–64
9. Gama, C.; Tenorio, R.; “Improvements for power systems performance: modeling, analysis
and benefits of TCSCs” ,Power Engineering Society Winter Meeting, 2000. IEEE Volume 2,
23–27 Jan. 2000 Page(s):1462 - 1467 vol.2
10. Xie Da; Niu Hui; Chen Chen; Wu Jishun, “An algorithm to control the power flow in large
systems based on TCSC “ ,Power System Technology, 1998. Proceedings. POWERCON ‘98.
Volume 1, 18–21 Aug. 1998 Page(s):344 - 348 vol.1
11. Geng Juncheng; Tong Luyuan; Ge Jun; Wang Zhonghong; “Mathematicalmodel for describing
characteristics of TCSC”, Power System Technology, 2002. Proceedings. PowerCon 2002.
Volume 3, 13–17 Oct. 2002 Page(s):1498 - 1502
12. Abdel-Moamen, M.A.; Narayana Prasad Padhy, “Power flow control and transmission loss
minimization model with TCSC for practical power networks”, Power Engineering Society
General Meeting, 2003,Volume 2, 13–17 July 2003
13. R. Mohan Mathur, Rajiv K. Varma; “ Thyristor-Based FACTS Controllers For Electrical
Transmission Systems”, A John Wiley & Sons, Inc. Publication,2002
14. Matsuki, J.; Ikeda, K.; Abe, M.; “Investigations of a thyristor-controlled series capacitor”,
Industrial Electronics, Control, and Instrumentation, 1996., Proceedings of the 1996 IEEE
IECON 22nd International Conference, Volume 2, 5–10 Aug. 1996 Page(s):683 – 688
15. Persson, J.; Rouco, L.; Soder, L “Linear analysis with two linear models of a thyristor-controlled
series capacitor”, Power Tech Conference Proceedings, 2003 IEEE Bologna, Volume 3,23–26
June 2003 Page(s):8pp
16. IEEE Power Engineering Society (1996) FACTS Application. IEEE press, New York
17. William D. Stevenson, Jr. “Elements of Power System Analysis”, McGraw-Hill Series in
Electrical Engineering, Fourth Edition,1982
18. Persson, J.; Rouco, L.; Soder, L ,”Linear analysis with two linear models of a thyristor-controlled
series capacitor”, Power Tech Conference Proceedings, 2003 IEEE Bologna
19. Rudra Pratap Singh, “Intoduction to MATLAB”,Ist Edition,2004
20. Chulin G, Luyuan T, Zhonghang W (2002) Stability Control of TCSC between interconnected
Power networks. In: Power system technology, 2002 proceedings, vol 3, pp 1943–1946
21. Chapman SJ (2004) MATLAB programming for engineers, 3rd edn. Thomsan Learning
22. Ally A, Rigby BS (2004) Member, IEEE, An investigation into the impact of a thyristor
controlle series capacitor-based closed loop power flow controller under fault conditions. IEEE
AFRICON, pp 675–681
23. Tan X, Tong L (1998) Characteristics and firing angle control of thyristor controlled series
compensation installations. In: IEEE Conference, pp 672–676
24. Billinton R, Fotuhi-Firuzaba M, Faried SO (1999) Power system reliability enhancement using
a thyristor controlled series capacitor. IEEE Trans Power Syst.14(1)
25. Xueqiang Z, Chen C (1998) Circuit Analysis of a thyristor controlled series compensation. In:
IEEE paper, pp 1067–1072
Chapter 37
Stockwell Transform and Hilbert
Transform Based Hybrid Algorithm
for Recognition of Power Quality
Disturbances
Ramesh Aseri and Ashwani Kumar Joshi
1 Introduction
Presently the electrical Power quality has become a major concern for both utilities
and customers because of the loads on the customer part get affected and might get
damaged. The utilities have opted the new open-access and competitive market power
policy. Now, the electricity consumers are in a unique position to demand a higher
quality of service. [2, 3]. The utilities and all other power supplier have to certify
a high quality service to remain competitive and Other issue which are responsible
for low power quality can be considered as: Voltage sag Voltage swell; Voltage
flicker; Voltage Interruption; Waveform Distortion; Voltage Fluctuation and Power
Frequency Variations A brief overview of the commonly occurred PQ disturbances,
their symptoms, possible causes and consequences are described in the Table 1.
The proposed methodology based on the combined features of Stockwell transform

and Hilbert transform for the recognition or detection of single stage and complex
PQ disturbances is detailed below step by step [4–7].
• First generate the voltage signal with PQ disturbance in the MATLAB using
mathematical relations.
• Decompose the voltage signal with PQ disturbance using the stockwell transform
to obtain the S-matrix.
R. Aseri · A. K. Joshi (B)

Department of Electrical Engineering, Jodhpur Institute of Engineeringand Technology, Jodhpur,
India
e-mail: ashwanijoshi27@gmail.com

https://doi.org/10.1007/978-981-15-8820-4_37
426 R. Aseri and A. K. Joshi
Table 1 Power Quality Standards

Standards Phenomena
IEEE St. 519-1992 Harmonic control in electrical power system
IEEE St. 1159-1995 Monitoring electric power quality
IEEE St. 1531-2003 Application and specification of harmonic filter
IEEE St. 1573-2003 Electronic power subsystem
IEEE St. 1566-2005 Adjustable speed drives rated 375 KW larger
IEEE St. 995-1987 Efficiency determination of alternating current adjustable speed drives
Table 2 Power Quality Disturbance and Their Quantification

Type of Categories Subcategory Typical Typical Typical
Disturbance Spectral Duration Voltage
Content Magnitude
Transient Impulsive Nanosecond 5 ns rise < 50 ns –
Microsecond 1 µs rise 50 ns to 1 ms
Millisecond 0.1 ms rise > 1 ms
Oscillatory LF <5 kHz 0.3–50 ms 0–4 pu
MF 5–500 kHz 20 µs 0–8 pu
HF 0.5–5 MHz 5 µs 0–4 pu
Short-Duration Instantaneous Sag (dip) – 0.5–30 cycles 0.1–0.9 pu
Variations Swells 0.5–30 cycles 1.1–1.8 pu
Momentary Interruption – 0.5 cycles-3 s < 0.1 pu
Sag 30 cycles-3 s 0.1–0.9 pu
Swells 30 cycles-3 s 1.1–1.4 pu
Temporary Interruption – 3 s to 1 min. < 0.1 pu
Sag 3 s to 1 min. 0.1–0.9 pu
Swells 3 s to 1 min. 1.1–1.2 pu
Long-Duration Interruption – – > 1 min. 0.0 pu
Variations Sustained
Under voltages – – > 1 min. 0.8–0.9 pu
Over voltages > 1 min. 1.1–1.2 pu
Voltage – – – Steady state 0.5–2%
Unbalance
Waveform D.C. Offset – – Steady State 1–0.1%
Distortion Harmonics – 0–100th Steady State 0–20%
Inter harmonics 0–6 kHz Steady State 0–2%
Notching – Steady State 0–1%
Noise Broadband Steady State
Voltage – – <25 Hz Intermittent 0.1–7%
Fluctuations
Power – – < 10 s –
Frequency
Variations
37 Stockwell Transform and Hilbert Transform … 427
Table 3 Different Type of Disturbance, Their Symptoms, Possible Causes and Consequences
Disturbance Type Symptom(s) Possible Cause(s) Consequence(s)
Interruption Complete loss of supply • Weather, storms, Affects all equipment
(exceeding 1 min) lightning and heavy
winds
• Accidents and
excavation
• Planned maintenance
• Line faults, blown
fuse
Overvoltage Long-term increase in • Light system loading Affects most
supply voltage (>+ 6%) • Poor regulation equipment without
Under voltage Long-term lowering of • Heavy, peak network internal backup
the supply voltage loading facilities
(<–10%) • Lack of VAR support
• Poor power factor
Momentary Short-term power loss • Circuit breaker
Interruption (200 ms-few seconds) tripping
• Fault clearing
• Bus transfer
Voltage Swell Medium-term (> • Circuit Capacitance Protection tripping or
200 ms) duration • Switching out large possible damage to
coupled with an loads tabulation and
increasing amplitude • Load rejection windings
(+10 to +30%) • Phase fault
Voltage Transient Short-duration • Lightning/capacitive Control resetting and
(sub-cycle) impulse switching major damage to
voltage/current spikes, • Low fault current trip sensitive electronic
large amplitude high protection components and
voltage gradients • Non-linear switching insulation
loads
• Transmitted noise
through the supply
system
Current Steady-state periodic • Increased use of Overheating of
Harmonics waveforms which non-linear circuit transformers and motor
deform the supply elements drives, increased
signal • High frequency power- loss and control
switches, TVs, command interference
computers and
fluorescent lighting
• Power factor
correction capacitor
• Negligent users
unaware of signal
pollution generated
by equipment
(continued)
Table 3 (continued)
Disturbance Type Symptom(s) Possible Cause(s) Consequence(s)
EMC Radiated and conducted • Incorrect wiring Interference with
(Electromagnetic interference, spurious • Common control signals,
Compatibility) signals which disturbances between spurious noise and
and proliferate on the supply and earth induced voltages
EMI Effects supply • Series disturbance
between supply and
neutral
• Generated by
unshielded electrical
equipment
• Unknown effects
regarding human
health matters
Flicker Series of systematic • Variable frequency Irritating light flicker
voltage fluctuations voltage variation, and control reset
fluorescent lights
• Erratic loads, reactive
power variation
Voltage Unbalance 3-phase load • Unbalanced Interruption of
interruptions, negative poly-phase loads, e.g. 3-phase operations
sequence and capacitor banks and
unsymmetrical voltages motors
• Obtain the sum absolute values of the absolute values of S-matrix.

• Decompose the voltage signal with PQ disturbance using the Hilbert transform.
• Obtain the absolute values of the output of Hilbert transform.
• Obtain power quality index by multiplying sum absolute values of the output of
S-matrix and absolute values of the output of the Hilbert transform. This power
quality index clearly detects and localizes the PQ disturbances.
• Plot the voltage signal and power quality index.
3 Recognition of Single Stage Power Quality Disturbances:

Simulation Results
Single stage PQ disturbances considered in the proposed study include pure sine
wave, voltage swell, voltage sag, momentary interruption, oscillatory transient,
impulsive transient, harmonics, notch and spike [8–12]. Analysis of above mentioned
PQ disturbances using the proposed Stockwell transform and Hilbert Transform
based algorithm is introduced in this chapter. Detailed analysis of the proposed or
planned PQ index to detect the various single stage PQ disturbances is provided in
the following subsection.
Fig. 1 Recognition of pure sinusoidal waveform of voltage signal a voltage signal without PQ
disturbance b proposed PQ index based on Stockwell transform and Hilbert transform
3.1 Pure Sinusoidal Voltage Signal
See Fig. 1.
3.2 Voltage Sag
See Fig. 2.
3.3 Voltage Swell
See Fig. 3.
Fig. 2 Recognition of voltage sag a voltage signal with PQ disturbance b proposed PQ index based
on Stockwell transform and Hilbert transform
3.4 Momentary Interruption
See Fig. 4.
3.5 Harmonics
See Fig. 5.
3.6 Flicker
See Fig. 6.
Fig. 3 Recognition of voltage swell a voltage signal with PQ disturbance b proposed PQ index
based on Stockwell transform and Hilbert transform
3.7 Notch
See Fig. 7.
3.8 Spike
See Fig. 8.
3.9 Oscillatory Transient
See Fig. 9.
Fig. 4 Recognition of momentary interruption a voltage signal with PQ disturbance b proposed

PQ index based on Stockwell transform and Hilbert transform
3.10 Impulsive Transient
See Fig. 10.
3.11 Classification of Single Stage PQ Disturbance
See Fig. 11.

The classification of the single stage PQ disturbances has been achieved using
the maximum values of the proposed PQ disturbances [13–15]. The maximum value
of the proposed PQ index for all the single stage PQ disturbances considered in the
proposed study is provided in the Table 4.
Based on the values of proposed PQ index, flowchart for the classification of
various single stage PQ disturbances is provided in Fig. 12. PQ disturbances have
been classified effectively using a set of if then else rules using the maximum values
of the proposed PQ index.
Fig. 5 Recognition of harmonics a voltage signal with PQ disturbance b proposed PQ index based
on Stockwell transform and Hilbert transform
4 Recognition of Complex Power Quality Disturbances:

Simulation Results
The simulation results related to detection of complex (multiple) PQ disturbances

using the proposed algorithm are described in this point. Complex PQ disturbances
considered in the proposed study include voltage sag with harmonics, voltage swell
with harmonics, momentary interruption with harmonics, Flicker and Harmonics,
Voltage Sag and Oscillatory Transient, Voltage Swell and Oscillatory Transient,
Momentary Interruption and Oscillatory Transient, Flicker and Oscillatory Tran-
sient, Harmonics and Oscillatory Transient, Voltage Sag and Impulsive Transient,
Voltage Swell and Impulsive Transient, Momentary Interruption and Impulsive Tran-
sient, Flicker and Impulsive Transient, Harmonics and Impulsive Transient, Voltage
Sag and Spike, Voltage Sag and Harmonics and Oscillatory Transient, Flicker and
Harmonics and Impulsive Transient, Voltage Sag, Oscillatory Transient, Harmonics
and Impulsive Transient. Analysis of above mentioned complex PQ disturbances
using the proposed Stockwell transform and Hilbert Transform based algorithm is
presented in this chapter [16–20]. Detailed analysis of the proposed PQ index to
detect the various complex PQ disturbances is provided in the following subsection.
Fig. 6 Recognition of flicker a voltage signal with PQ disturbance b proposed PQ index based on
Stockwell transform and Hilbert transform
4.1 Voltage Sag with Harmonics
See Fig. 11.
4.2 Momentary Interruption and Harmonics
See Fig. 12.
4.3 Flicker and Harmonics
See Fig. 13.

Fig. 7 Recognition of notch a voltage signal with PQ disturbance b proposed PQ index based on
4.4 Voltage Swell and Oscillatory Transient
See Fig. 14.
4.5 Momentary Interruption and Oscillatory Transient
See Fig. 15.
4.6 Flicker and Oscillatory Transient
See Fig. 16.

Fig. 8 Recognition of spike a voltage signal with PQ disturbance b proposed PQ index based on
4.7 Harmonics and Oscillatory Transient
See Fig. 17.
4.8 Voltage Sag and Impulsive Transient
See Fig. 18.
4.9 Voltage Swell and Impulsive Transient
See Fig. 19.

Fig. 9 Recognition of oscillatory transient a voltage signal with PQ disturbance b proposed PQ

index based on Stockwell transform and Hilbert transform
4.9.1 Classification of Complex Power Quality Disturbances
The classification of complex stage PQ disturbances has been achieved using the
maximum values of the proposed PQ disturbances [16–20]. The maximum value
of the proposed PQ index for all the complex PQ disturbances considered in the
proposed study is provided in the Table 3. Based on the values of the proposed PQ
index, flowchart for the classification of various complex PQ disturbances is provided
in Fig. 4. PQ disturbances have been classified effectively using a set of if then else
rules using the maximum values of the proposed PQ index.
5 Conclusions
Detection of Single Stage Power Quality Disturbances

An algorithm based on the combined features of Stockwell transform and Hilbert
transform has been proposed for the recognition of single stage PQ disturbances.
Voltage signal with single stage PQ disturbance has been decomposed using the
Stockwell transform to obtain the S-matrix. Sum absolute values of the absolute
Fig. 10 Recognition of impulsive transient a voltage signal with PQ disturbance b proposed PQ

index based on Stockwell transform and Hilbert transform
values of S-matrix are obtained. Further, the voltage signal with PQ disturbance
is also decomposed using the Hilbert transform and absolute values of the output
of Hilbert transform is obtained. Proposed PQ index is obtained by multiplying
sum absolute values of the output of S-matrix and absolute values of output of the
Hilbert transform. This PQ index clearly detects and localizes the PQ disturbances.
Classification of the PQ disturbances has been achieved using the maximum values
of proposed PQ index. Investigated PQ disturbances include voltage sag, voltage
swell, momentary interruption, oscillatory transient, impulsive transient, harmonics,
notch and spike. This has been concluded that proposed approach based on combined
features of the Stockwell Transform and Hilbert Transform is found to be effective in
the recognition of the single stage PQ disturbances. This algorithm is more effective
compared to the algorithm based on the discrete wavelet transform.
Detection of Complex Power Quality Disturbances
Proposed algorithm based on the combined features of Stockwell transform and
Hilbert transform has also been tested for recognition of complex PQ disturbances.
Voltage signal with complex PQ disturbance has been decomposed using the Stock-
well transform to obtain S-matrix. Sum absolute values of the absolute values of
S-matrix are obtained. Further, the voltage signal with complex PQ disturbance is
also decomposed using the Hilbert transform and absolute values of output of Hilbert
Fig. 11 Flowchart for classification of single stage PQ disturbances
Table 4 Maximum values of

Sr. No. Operational Event Maximum value of PQI
Proposed PQ Index For
Magnitude
Single Stage PQ Disturbances
1 Pure sine wave 3.9895
2 Voltage sag 3.99
3 Voltage swell 4.0437
4 Momentary Interruption 3.8951
5 Harmonics 11.2380
6 Flicker 8.2938
7 Notch 15.5470
8 Spike 19.3988
9 Oscillatory Transient 148.0551
10 Impulsive Transient 304.4628
transform is obtained. Proposed PQ index is obtained by multiplying sum absolute

values of output of S-matrix and absolute values of output of the Hilbert transform.
This PQ index clearly detects and localizes the complex PQ disturbances. Classifi-
cation of complex PQ disturbances has been achieved using the maximum values of
proposed PQ index. Investigated complex PQ disturbances include voltage sag with
Table 5 Maximum Values Of Proposed Power Quality Index For Complex PQ Disturbances
Sr. No. Complex PQ Disturbance Maximum value of PQI Magnitude
1 Voltage Sag with Harmonics 11.2121
2 Voltage Swell and Harmonics 13.2071
3 Momentary Interruption and Harmonics 09.1619
4 Flicker and Harmonics 12.2506
5 Voltage Sag and Oscillatory Transient 142.6547
6 Voltage Swell and Oscillatory Transient 148.6776
7 Momentary Interruption and Oscillatory 138.9654
Transient
8 Flicker and Oscillatory Transient 145.7373
9 Harmonics and Oscillatory Transient 137.9313
10 Voltage Sag and Impulsive Transient 304.8193
11 Voltage Swell and Impulsive Transient 310.1465
12 Momentary Interruption and Impulsive 300.7539
Transient
13 Flicker and Impulsive Transient 285.6506
14 Harmonics and Impulsive Transient 289.0105
15 Voltage Sag and Spike 19.3822
16 Voltage Sag, Harmonics and Oscillatory 135.5474
Transient
17 Flicker, Harmonics and Impulsive 271.0759
Transient
18 Voltage Sag, Oscillatory Transient, 161.7251
Harmonics and Impulsive Transient
harmonics, voltage swell with harmonics, momentary interruption with harmonics,

Flicker and Harmonics, Voltage Sag and Oscillatory Transient, Voltage Swell and
Oscillatory Transient, Momentary Interruption and Oscillatory Transient, Flicker
and Oscillatory Transient, Harmonics and Oscillatory Transient, Voltage Sag and
Impulsive Transient, Voltage Swell and Impulsive Transient, Momentary Interrup-
tion and Impulsive Transient, Flicker and Impulsive Transient, Harmonics and Impul-
sive Transient, Voltage Sag and Spike, Voltage Sag and Harmonics and Oscillatory
Transient, Flicker and Harmonics and Impulsive Transient, Voltage Sag, Oscilla-
tory Transient, Harmonics and Impulsive Transient. This has been concluded that
proposed approach based on combined features of the Stockwell Transform and
Hilbert Transform is found to be effective in the recognition of the complex PQ
disturbances. This algorithm is more effective compared to the algorithm based on
the discrete wavelet transform.
Fig. 12 Recognition of simultaneous occurrence of voltage sag and harmonics a voltage signal
with PQ disturbance b proposed PQ index based on Stockwell transform and Hilbert transform
Fig. 13 Recognition of simultaneous occurrence of voltage swell and harmonics a voltage signal
Fig. 14 Recognition of simultaneous occurrence of momentary interruption and harmonics

a voltage signal with PQ disturbance b proposed PQ index based on Stockwell transform and
Hilbert transform
Fig. 15 Recognition of simultaneous occurrence of flicker and harmonics a voltage signal with
PQ disturbance b proposed PQ index based on Stockwell transform and Hilbert transform
Fig. 16 Recognition of simultaneous occurrence of voltage swell and oscillatory transient a voltage
signal with PQ disturbance b proposed PQ index based on Stockwell transform and Hilbert transform
Fig. 17 Recognition of simultaneous occurrence of momentary interruption and oscillatory tran-

sient a voltage signal with PQ disturbance b proposed PQ index based on Stockwell transform and
Hilbert transform
Fig. 18 Recognition of simultaneous occurrence of flicker and oscillatory transient a voltage signal
Fig. 19 Recognition of simultaneous occurrence of harmonics and oscillatory transient a voltage

Fig. 20 Recognition of simultaneous occurrence of voltage sag and impulsive transient a voltage
Fig. 21 Recognition of simultaneous occurrence of voltage swell and impulsive transient a voltage
Fig. 22 Flowchart for classification of complex stage PQ disturbances

Appendix A
STOCKWELL TRANSFORM PROGRAMMING

function ST=stran(h)
% COMPUTE S-TRANSFORM WITHOUT FOR LOOPS
[~,N]=size(h); % h is a 1xN one-dimensional series
nhaf=fix(N/2);
odvn=1;
if nhaf*2==N;
odvn=0;
end
f=[0:nhaf -nhaf+1-odvn:-1]/N;
Hft=fft(h);
%Compute all frequency domain Gaussians as one matrix
invfk=[1./f(2:nhaf+1)]';
W=2*pi*repmat(f,nhaf,1).*repmat(invfk,1,N);
G=exp((-W.^2)/2); %Gaussian in freq domain
% End of frequency domain Gaussian computation
% Compute Toeplitz matrix with the shifted fft(h)
HW=toeplitz(Hft(1:nhaf+1)',Hft);
% Exclude the first row, corresponding to zero frequency
HW=[HW(2:nhaf+1,:)];
% Compute Stockwell Transform
ST=ifft(HW.*G,[],2); %Compute voice
%Add the zero freq row
st0=mean(h)*ones(1,N);
ST=[st0;ST];
end
References
1. Lima MAA, Cerqueira AS, Courya DV, Duqueb CA (2012) A novel method for power quality
multiple disturbance decomposition based on Independent Component Analysis. International
Journal of Electrical Power and Energy Systems, vol. 42, pp 593–604
2. Saurabh Kamble and Ishita Dupare, “Detection of power quality disturbances using Wavelet
Transform and artificial neural network,” International Conference on Magnetics, Machines
and Drives, 2014
3. Rahul Dubey, S. R.Samantaray, B. Chitti Babu and S. Nandha Kumar, “Detection of power
quality disturbances in presence of DFIG wind farm using Wavelet Transform based energy
function,” IEEE International Conference, 2011
4. Norman C.F.Tse, John Y.C.Chan, Wing-Hong Lau and Loi Lei Lai, “Hybrid Wavelet and
Hilbert Transform with frequency-shifting decomposition for power quality analysis,” IEEE
Transactions on Instrumentation and Measurement, vol. 61, no. 12, pp-3225–3233, Dec 2012
5. Mahela OP, Shaik AG, Gupta N (2015) A critical review of detection and classification of
power quality events. Renew Sustain Energy Rev 41:495–505
6. Tse NCF, Chan JYC, Lau W-H, Lai LL (2012) Hybrid wavelet and hilbert transform with
frequency-shifting decomposition for power quality analysis. Instrumentation and Measure-
ment, IEEE Transactions on 61(12):3225–3233
7. Stockwell RG, Mansinha L, Lowe RP (1996) Localization of the complex spectrum: the s
transform. Signal Processing, IEEE Transactions on 44(4):998–1001
8. Eristi H, Yildirim O, Eristi B, Demir Y (2014) Automatic recognition system of underlying
causes of power quality disturbances based on s-transform and extreme learning machine.
International Journal of Electrical Power Energy Systems 61:553–562
9. Biswal M, Dash PK (2013) Detection and characterization of multiple power quality
disturbances with a fast s-transform and decision tree based classifier. Digit Signal Proc
23(4):1071–1083
10. Kaddah SS, Abo-Al-Ez KM, Megahed TF, Osman MG (2016) Probabilistic power quality
indices for electric grids with increased penetration level of wind power generation. Interna-
tional Journal of Electrical Power Energy Systems 77:50–58
11. Ray PK, Mohanty SR, Kishor N (2011) Disturbance detection in grid-connected distributed
generation system using wavelet and s-transform. Electr Power Syst Res 81(3):805–819
12. C.-T. Hsu, R. Korimara, and T.-J. Cheng, “Power quality analysis for the distribution systems
with a wind power generation system,” Computers Electrical Engineering, pp. –, 2015
13. E.A. Cano Plataa and H.E. Tacca, “Power load identification,” Journal of the Franklin Institute,
vol. 342, pp 97–113, AUG2004
14. Akhbardeha Alireza, Junnilaa Sakari, Koivuluomaa Mikko, Koivistoinenb Teemu (Oct 2007)
Vaino¨ Turjanmaab, Tiit Koobib and Alpo Varria, “Towards a heart disease diagnosing system
based on force sensitive chair’s measurement, biorthogonal wavelets and neural networks”.
Eng Appl Artif Intell 20:493–502
15. Moravej Zahra (2015) Jamal Dehghani Ashkezari and Mohammad Pazoki, “An effective
combined method for symmetrical faults identification during power swing”. Electrical Power
and Energy Systems 64:24–34
16. Bhim Singh Sunil Kumar Dube and Sabha Raj Arya, “An improved control algorithm of
DSTATCOM for power quality improvement,” Electrical Power and Energy Systems, vol.
64,pp 493–504, AUGUST 2015
17. Sundarabalan CK, Selvi K (2015) Compensation of voltage disturbances using PEMFC
supported Dynamic Voltage Restorer. Electrical Power and Energy Systems 71:77–92
18. Om Prakash Mahela and Abdul Gafoor Shaik, “Power quality improvement in distribution
network using DSTATCOM with battery energy storage system,” Electrical Power and Energy
Systems, vol. 83, pp 229–240, APRIL 2016
19. C.K. Sundarabalan and K. Selvi, “PEM fuel cell supported distribution static compensator for
power quality enhancement in three-phase four-wire distribution system,” international journal
of hydrogen energy, vol. 39, pp 19051–19066, OCT. 2014
20. N. Srinivasa Raoa, Dr. A. Selwin Mich Priyadharsonb and Dr. J. Praveen, “Simulation of Arti-
ficial Intelligent Controller based DVR for Power Quality Improvement,” Procedia Computer
Science, vol.47, pp 153 – 167, july 2015
Chapter 38
Detection and Classification
of Transmission Line Faults Using
Combined Features of Stockwell
Transform, Hilbert Transform,
and Wigner Distribution Function
Tanmay Bhati and Harish Kumar Khyani
1 Introduction
Introduction mainly focuses on the detection and classification of transmission line

faults using combined features of the Stockwell transform, Hilbert transform, and
Wigner Distribution features. Timely detection of fault is main aspect of any kind
of High-Tension lines weather it is of transmission system or distribution system.
Detection and classification subjects are to be studied in details to avoid repetitive
and dummy trappings of system. Each time when a system exhibits a fault it creates
jerk on equipment’s and affects stability as well. All this factor is necessary to design
a precise relay system looking to the study parameters [1].
1.1 Types of Faults
Following are the faults generally take place in power system:

• Over current: It occurs mainly due to short-circuit or leakage or due to corona
effect and sometimes due to overload on the supply system.
• Under voltage: It mainly occurs either on the failure of alternator’s field or short
circuits because of more voltage drop in machines and lines.
• Unbalance: It occurs either on two phases or breaking of one of the conductors
or grounding of one or on a short circuit of two phases. In such type of condition,
T. Bhati (B) · H. K. Khyani

Department of Electrical Engineering, Jodhpur Institute of Engineering & Technology, Jodhpur,
India
e-mail: tanmaybhati@gmail.com

https://doi.org/10.1007/978-981-15-8820-4_38
450 T. Bhati and H. K. Khyani
different current flow through different phases and fault is known as an unbalanced
fault.
• Reversed power: This fault occurs only in interconnected systems. On failure
of the field, a generator starts working as a motor and takes power instead of
delivering the power means the flow of power is reversed. Similarly in case of
feeders which are connected in parallel, if some fault occurs on any one of the
feeders, then the fault is fed from both the ends, means again direction of flow of
power is reversed in case of the faulty feeder.
• Surges: Whenever a severe fault occurs or lightning takes place in the neighboring
circuits, then in the lines, some short-lived waves of very high voltages and currents
are setup. It may be considered as a high voltage of high frequency and such fault
is known as the surge.
• Short-circuit currents
When the current gets least resistive path or non-resistive path a very heavy current
flow through this least resistive path or non-resistive path. This phenomenon is known
as short-circuiting. The heavy current is known as short-circuit current. This short-
circuit current is very harmful to the equipment. It is necessary to disconnect the faulty
section to save the equipment from complete interruption of supply and damage for
which protective gear such as fuses, circuit of suitable capacity are required. The
fault in three-phase power system may occur:
• Single phase to earth,
• Phase to phase,
• Two phases to earth,
• Phase to phase and at the same time from the third phase to earth,
• All the three phases to earth,
• All the three phases short-circuited.
The first four types of fault are called an unsymmetrical fault. While the latter two
types of fault are called a symmetrical fault. Although the symmetrical fault gives
the more severe duty on the circuit breaker and the unsymmetrical faults are more
prevalent. The calculations of a symmetrical short circuit are very important for the
purpose of determination of circuit breaker ratings.
Various Signal Processing Techniques like wavelet transform (WT), artificial
neural network (ANN), Fuzzy logic (FL) etc. are used for the identification of faults
and its classification for the protection system [2]. In this context, a comprehensive
survey is performed to review conceptually, intelligent fault detection and classifi-
cation techniques that are worked upon for diagnosis of fault on transmission lines
[3]. A simplified fault locating approach which can be used for multi-terminal trans-
mission lines using unsynchronized measurements is proposed by the author in [4].
Procedures for data synchronization are deployed for spotting/identifying faulted
phase before location of fault location has been initiated. Fault location algorithm
will not depend on variations in the fault resistance and variations source impedance.
Location algorithm and faulted phase is effectively tested for all types of faults and
various high fault resistances have been used. Result states that newly proposed
38 Detection and Classification of Transmission Line … 451
multi-terminal fault location algorithm is not only fast and accurate but also immune
to power system transients. In [5] authors proposed a method supported by WT fort
identification, classification and location of faults on power transmission lines. A
Global Positioning System clock is being used to synchronize sampling of voltage
and current signals at both the ends of the transmission line. The detailed coefficients
of current signals of both the ends are employed or utilized to calculate fault indices.
These resultant fault indices are compared with threshold values for the detection and
classification of the faults. In order to locate the fault (or to identify the location of
fault), Artificial Neural Networks are being employed, which uses the approximate
decompositions of the voltages and currents of the local end [6]. Accurate estima-
tion of fault location on transmission line is required for inspection of line, mainte-
nance of line, and rectification of fault incident on line. Hence, for protection relay
functioning, the identification of fault location is really very important subjects in
power system [7]. Various mathematical and signal processing techniques have been
used for recognition and classification of faults on transmission. Yadav and Sweta-
padma [8] discussed a fault detection and classification algorithm which effectively
protects the transmission line against the faults is proposed. Algorithm is based on
the combination of Wavelet transform (WT) and linear discriminant analysis (LDA).
2 Proposed Test System and Algorithm for Recognition

of Faults on Transmission Line
In this section, the proposed test transmission line used for the study of detection
and classification of transmission line faults is detailed. The methodology based on
combined features of Stockwell transform, Hilbert transform, and Wigner Distribu-
tion Function has been illustrated for implementation using current signals. A brief
description of the Stockwell transform, Hilbert transform, and Wigner Distribution
Function is also presented.
2.1 Stockwell’s Transform
There are some different methods of achieving the S-transform. At this point, the
relationship between S-transform and STFT is introduced and the type of S-transform
is derived from the “phase correction” of the CWT. The short-time Fourier transform
of a signal, i.e., h(t) is defined by means of the following relation:
+∞
ST F T (τ, f ) = ∫ h(t)g(τ − t)e− j2π f t dt (1)
−∞
where τ and f denote the time of spectral localization and Fourier frequency, respec-
tively, and window function is denoted by g(t). This S-transform may be derived with
the help of the above equation by replacing g(t) i.e. the window function by means
of the Gaussian function which is shown below.

|f| f 2t2
g(t) = √ e− 2 (2)
2π
Then the S-transform is defined as

+∞ | f | f 2 (τ −t)2
S(τ, f ) = ∫ h(t) √ e 2 e− j2π f t dt (3)
−∞ 2π
Hence, the S-transform is a special case of STFT with a Gaussian window function.
If the window of S-transform is wider in time domain, S-transform can provide better
frequency resolution for lower frequency. While the window is narrower, it can
provide better time resolution for higher frequency. The output of the S-transform is
a matrix known as S-matrix. The information related to the frequency and amplitude
of the signal can be derived from the S-matrix.
2.2 Hilbert Transform
The main objective of the Hilbert Transform is to express or demonstrate a substi-

tute/alternate process to present spectral analysis tools intended to provide the time–
frequency–energy depiction of time series data. Also, the method attempts to describe
non-stationary data locally. In order to calculate instantaneous frequencies as well as
amplitudes and explain the signal more locally, Hilbert transform was used rather than
a wavelet-based transform or Fourier transform. The below given equation explains
the Hilbert transform, which can be explained/written for any function, i.e., x(t) of
LP class. PV denotes Cauchy’s principal value integral.
1 ∞ x(τ )
H [x(t)] = PV ∫ dτ (4)
π −∞ t − τ
Not all functions give “good” Hilbert transforms, meaning those which produce
physical instantaneous frequencies. For example, functions with nonzero means will
give negative frequency contributions using the Hilbert transform. Therefore, the
signals which can be analyzed using the Hilbert transform must be restricted so
that their calculated instantaneous frequency functions have a physical connotation.
Subsequently, it describes the empirical mode decomposition. Essentially, it is an
algorithm which decomposes nearly every signal to a finite/limited set of functions,
which contain “good quality” Hilbert transforms that create physically meaningful
instantaneous frequencies.
2.3 Wigner Distribution Function
The Wigner distribution function (WDF) is utilized in signal processing to trans-

form in time–frequency analysis. The Wigner distribution function was initially
proposed by Eugene Wigner in physics to account for quantum corrections toward
classical statistical mechanics in the year 1932, and it is of much significance
in quantum mechanics in phase space (Wigner quasi-probability distribution is also
called the Wigner function or the Wigner–Ville distribution). The shared algebraic
structure between time–frequency conjugate pairs and position–momentum is spec-
ified, it too beneficially serves in signal processing, as a change/transform in time–
frequency analysis, which is the main subject of this article. In comparison with
a short-time Fourier transform such as the Gabor transform, the Wigner distribution
function which will be able to furnish superior lucidity in some cases. The Gabor
transform is given by the following mathematical relation:
∞
G x (t, f ) = ∫ e−π(τ −t) e− j2π f τ x(τ )dτ
2
(5)
−∞
The Wigner distribution function is given as

∞ τ ∗ τ − j2π f τ
Wx (t, f ) = ∫ x t + x t− e dτ (6)
−∞ 2 2
2.4 Proposed Transmission Line Used as Test System
The proposed test transmission line is connected between two buses of the complex
power system network. The large area utility on both ends of the transmission line
is represented by generators 1 and 2 (G1 and G2). The test system of transmission
line used for the study is illustrated in Fig. 1. Total line length is taken as 230 km.
The line is modeled in four L1, L2, L3, and L4 each having a length of 57.5 km.
Fault is created at points F1, F2, and F3. The fault location points F1 and F3 are used
to investigate the effect of fault location whereas fault location point F2 situated at
the middle of the line is used in all other cases. The measurement of the voltage
Fig. 1 Test transmission line

Table 1 Technical
Particulars of parameters Values of parameters
parameters of test system [5]
Voltage rating of system 500 kV
Frequency 60 Hz
Source resistance (ohm) 17.77
Source inductance 0.1218 H
Positive sequence resistance of line 0.01273 /km
Zero sequence resistance of line 0.3864 /km
Positive sequence inductance of line 0.9337e-3 H/km
Zero sequence inductance of line 4.1264e-3 H/km
Positive sequence capacitance of line 12.74e-9 F/km
Zero sequence capacitance of line 7.751e-9 F/km
and current is carried out using the three-phase VI measurement block of Simulink
(Table 1).
2.5 Proposed Methodology
The methodology proposed for detection and classification of transmission line faults
to design an effective transmission line protection scheme is designed for the current-
based protection scheme used the Stockwell Transform, Hilbert Transform, and
Wigner Distribution Function for processing of the signals.
2.5.1 Protection Scheme Using Current Signal
Current signal based algorithm is proposed for detection and classification of faults
on the transmission line to design an effective transmission line protection scheme
consisting of the design with the following steps detailed below.
• To start with (Initially), the current signal is decomposed by means of Stockwell
transform with a sampling frequency of 3.2 kHz to gain S-matrix. Further, S-
matrix is converted to a matrix having the absolute values corresponding to each
element of S-matrix. Median of this new matrix is calculated and designated as
ST-index.
• Secondly, the current signal is decomposed using Hilbert transform with a
sampling frequency of 3.2 kHz. Absolute values output so obtained are designated
as H-index.
• Thirdly, the current signal is decomposed using Wigner distribution function with
a sampling frequency of 3.2 kHz. Absolute values obtained at the first level of
decomposition are designated as WDF-index.
• Finally, element by element product of ST-index, H-index, and DWF-index is

obtained and designated as Fault index. A threshold value of 5000 is selected to
distinguish the faulty phase from the healthy phase as well as the faulty event
from the healthy event.
• The proposed algorithm has been implemented under MATLAB/Simulink envi-
ronment.
3 Detection and Classification of Transmission Line Faults

Using Current-Based Fault Index
Simulation results related to detection and classification of transmission line faults

using combined features of Stockwell Transform, Hilbert Transform, and Wigner
Distribution Function based approach are detailed in this section. The results related
to case studies such as reverse power flow, effects of fault location, switching of
inductive load, and switching of capacitive load have been described in this section.
3.1 Healthy Condition
It is observed from Fig. 2b that the proposed ST-index has zero values throughout
the range. Figure 2c depicts that the value of the proposed H-index is constant over
the entire time range. It is observed from Fig. 2d that the value of WDF-index is
zero except at the central location where it is non-zero representing the frequency
distribution of fundamental frequency. The proposed fault index has zero values over
entire time range as depicted in Fig. 2e. These plots are taken as reference plots for
detection of faulty events.
3.2 Line to Ground Fault
This can be observed from Fig. 3a current in Phase-A increases after the faulty event.
It is observed from Fig. 3b that the proposed ST-index has zero values throughout
the range corresponding to Phases-B and C whereas these values corresponding to
Phase-A are high indicating that there is a fault on this phase. Figure 3c indicates that
the value of proposed H-index is constant over the entire time range for the Phases-B
and C whereas it has high values for Phase-A indicating the faulty phase.
Fig. 2 Detection of transmission line faults using hybrid algorithm based on combined features
of Stockwell transform, Hilbert transform, and Wigner distribution function results for healthy
condition a current waveform b ST-index c H-index d WDF-index e proposed fault index
Figure 3d indicates that the value of proposed WDF-index has very high values
corresponding to Phase-A near the central location of time whereas these values are
low for Phases-B and C. It is depicted from Fig. 3e that proposed fault index corre-
sponding to Phase-A has higher values above the threshold indicating the presence of
a fault. The values of the fault index corresponding to Phases-B and C are below the
threshold and approximately equal to zero indicating the healthy phases. Hence, LG
fault on Phase-A has been successfully detected using the proposed algorithm based
on Stockwell Transform, Hilbert Transform, and Wigner Distribution Function.
Similarly, in the rest of the section the results are shown through the figure from
which we can analyze the result of different fault conditions.
Fig. 3 Detection of transmission line faults using hybrid algorithm based on combined features of
Stockwell transform, Hilbert transform, and Wigner distribution function results for line to ground
fault a current waveform b ST-index c H-index d WDF-index e proposed fault index
3.3 Double Line Fault and Double Line to Ground Fault

(LLG)
458
Fig. 4 Detection of transmission line faults using hybrid algorithm based on combined features of Stockwell transform, Hilbert transform, and Wigner distribution
function results for double line fault and double line to ground fault a current waveform, b ST-index, c H-index, d WDF-index, e proposed fault index
T. Bhati and H. K. Khyani
3.4 Three Phase to Ground Fault
Fig. 5 Detection of transmission line faults using hybrid algorithm based on combined features of
Stockwell transform, Hilbert transform, and Wigner distribution function results for three phase to
ground fault a current waveform, b ST-index, c H-index, d WDF-index, e proposed fault index
Fig. 6 Effect of fault location on detection of transmission line faults using hybrid algorithm based
on combined features of Stockwell transform, Hilbert transform, and Wigner distribution function
results for line to ground fault at 25% of line length a current waveform, b ST-index c H-index,
d WDF-index, e proposed fault index
3.5 Effect of Fault Location
To generalize the proposed algorithm, it is tested at different locations of fault on the

transmission line. All the faults have been tested at 25 and 75% of line length from
the sending end of the line. However, the results of LG fault are shown as 25 and
75% of line length in the following sections. The results have been discussed for all
the phases at 50% of line length in the above sections.
3.5.1 Line to Ground Fault at 25% Line Length
This can be observed from Fig. 6a that current in Phase-A increases after the
faulty event. It is observed from Fig. 6b that the proposed ST-index has zero values
throughout the range corresponding to Phases-B and C whereas these values corre-
sponding to Phase-A are high indicating that there is a fault on this phase. Figure 6c
indicates that the value of proposed H-index is constant over the entire time range for
the Phases-B and C whereas it has high values for the Phase-A indicating the faulty
phase. Figure 6d indicates that the value of proposed WDF-index has very high values
sponding to Phase-A has higher values above the threshold indicating the presence
of fault. The values of fault index corresponding to Phases-B and C are below the
fault on Phase-A at a line length of 25% from sending end of the transmission line
has been successfully detected using the proposed algorithm based on Stockwell
Transform, Hilbert Transform, and Wigner Distribution Function.
3.5.2 Line to Ground Fault at 75% Line Length
Fig. 7 Effect of fault location on detection of transmission line faults using hybrid algorithm based
on combined features of Stockwell transform, Hilbert transform and Wigner distribution function
results for line to ground fault at 75% of line length a current waveform, b ST-index, c H-index,
d WDF-index, e proposed fault index
Phases-B and C whereas it has high values for Phase-A indicating the faulty phase.
of fault. The values of fault index corresponding to Phases-B and C are below the
fault on Phase-A at a line length of 75% from sending end of the transmission line
has been successfully detected using the proposed algorithm based on Stockwell
Transform, Hilbert Transform, and Wigner Distribution Function.
3.6 Effect of Reverse Power Flow
Phases-B and C whereas it has high values for Phase-A indicating the faulty phase.
of a fault.
The values of the fault index corresponding to phases-B and C are below the
fault on Phase-A at line length of 50% from sending end of the transmission line has
been successfully detected using the proposed algorithm based on Stockwell Trans-
form, Hilbert Transform, and Wigner Distribution Function during the condition of
reverse power flow on the transmission line.
3.7 Effect of Switching Transients
To generalize the proposed algorithm, it is tested for the switching transients. The
results related to the switching of inductive load and capacitive load have been
discussed in the following sections.
Fig. 8 Effect of reverse power flow on detection of transmission line faults using hybrid algorithm
based on combined features of Stockwell transform, Hilbert transform and Wigner distribution
function results for line to ground fault at 50% of line length a current waveform, b ST-index,
c H-index, d WDF-index, e proposed fault index
3.7.1 Switching of Capacitive Load
This can be observed from Fig. 9a that current contains switching transients due
to switching the capacitive load. It is observed from Fig. 9b that the proposed ST-
index has zero values before switching on the capacitive load and finite values at the
time of switching the capacitor. Figure 9c indicates that the value of the proposed
H-index is constant before switching on the capacitive load and finite values at the
time of switching the capacitor. Figure 9d indicates that the value of proposed WDF-
index has high values corresponding to all phases near the central location of time.
It is depicted from Fig. 9e that proposed fault index corresponding to all phases has
higher values but below the threshold indicating that the event is switching and not
the faulty. Hence, switching of the capacitive load has been differentiated from the
faulty events using the proposed algorithm based on Stockwell Transform, Hilbert
Transform, and Wigner Distribution Function.
Fig. 9 Simulation results using hybrid algorithm based on combined features of Stockwell trans-
form, Hilbert transform, and Wigner distribution function for the effect of capacitive load switching
a current waveform, b ST-index, c H-index, d WDF-index, e proposed fault index
3.8 Switching of Inductive Load
This can be observed from Fig. 10a that current in all phases increases due to increased
load after switching the inductive load. It is observed from Fig. 10b that the proposed
ST-index has zero values before switching on the inductive load and finite values at
the time of switching the load. Figure 10c indicates that the value of the proposed
H-index is constant before switching on the inductive load and finite values at the
time of switching the load. Figure 10d indicates that the value of the proposed WDF-
index has high values corresponding to all phases near the central location of time.
It is depicted from Fig. 10e that proposed fault index corresponding to all phases has
higher values but below the threshold indicating that the event is switching and not
the faulty. Hence, switching of the inductive load has been differentiated from the
faulty events using the proposed algorithm based on Stockwell Transform, Hilbert
Transform, and Wigner Distribution Function.
Fig. 10 Simulation results using hybrid algorithm based on combined features of Stockwell trans-
form, Hilbert Transform and Wigner Distribution Function for the effect of inductive load a current
waveform b ST-index c H-index d WDF-index e proposed fault index
3.9 Classification of Faults
The peak values of the proposed fault index based on the current-based features are
provided in Table 2. It can be observed that if values of fault index are below 5000
for all the phases then there is no fault on the system and this indicates the healthy
condition. If there are only values of fault index is above 5000 corresponding to one
Table 2 Peak values of current-based fault index

Sr. No. Type of fault Peak values of fault index
Phase-A Phase-B Phase-C
1 Healthy condition 0 0 0
2 Line to ground fault 8000 100 100
3 Double line fault 4 × 104 3 × 104 100
4 Double line to ground fault 2.5 × 104 3.5 × 104 100
5 Three-phase fault involving ground 4.5 × 104 3 × 104 4 × 104
phase, then this indicates the presence of LG fault. If the fault index corresponding to
two phases is above the threshold then this indicates the presence of LL or LLG fault.
These can be discriminated from each other by using certain ground fault index. If
the fault index corresponding to all the three phases is above the threshold then this
indicates the presence of LLLG fault.
4 Conclusions
In this paper, an accurate protection system against various types of power system
faults like line to ground, double line, double line to ground, and three-phase faults
including ground three phase to ground have been developed. An algorithm based on
combined features of Stockwell transform, Hilbert transform, and Wigner distribu-
tion function has been proposed for the detection and classification of transmission
line faults to design effective protection scheme.
The method is tested on a test transmission line for different cases which
includes change in line length, reverse power flow condition and switching tran-
sients using MATLAB/Simulink. The advantage of this method is its effectiveness
for different cases where many traditional algorithms have failed. From the simula-
tion results, it has been found that the normal value of fault index in healthy condition
stays zero. In case of faulty event the value of fault index varies between 6.121e+03 as
minimum value and 45.36e+03 as maximum value in case of current-based analysis
when threshold is 5000. In case of voltage-based analysis, the value of fault index
varies between 1.290e+03 as minimum value and 9.101e+3 as maximum value when
threshold is 1000. Hence, there exist an accurate difference between a healthy and a
faulty condition.
References
1. Yang L, Xiao H, Zidong W, Dong-Hua X (2015) Fault Detection and Diagnosis for a Class of
Nonlinear Systems with Decentralized Event-triggered Transmissions. IFAC-Pap Line (Elsevier)
48–21:1134–1139
2. Durga Prasad Ch, Srinivasu N (2015) Fault detection in transmission lines using instantaneous
power with ED based fault index. Proceedia Technol 21:132–138
3. Ferreira VH, Zanghi R, Fortes MZ, Sotelo GG, Silva RBM, Souza JCS, Guimarães CHC, Gomes
Jr.Fsjjglds S (2016) A survey on intelligent system application to fault diagnosis in electric power
system transmission lines Elect. Power Syst Res 136:135–153
4. Hussain S, Osman AH (2016) Fault location scheme for multi-terminal transmission lines using
unsynchronized measurements. Int J Elect Power Energy Syst 78:277–284
5. Gafoor Shaik A, Pulipaka RRV (2015) A new wavelet based fault detection, classification and
location in transmission lines. Int J Elect Power Energy Syst 64:35–40
6. Dehghani M, Khooban MH, Niknam T (2016) Fast fault detection and classification based on
a combination of wavelet singular entropy theory and fuzzy logic in distribution lines in the
presence of distributed generations. Int J Elect Power Energy Syst 78:455–462
7. Morais AP, Cardoso Júnior G, Mariotto L, Marchesan G (2016) Fault location scheme for multi-
terminal transmission lines using unsynchronized measurements. Int. J. Elect. Power Energy
Syst 78:277–284
8. Swetapadma A, Yadav A (2016) Directional relaying using support vector machine for double
circuit transmission lines including cross-country and inter-circuit faults. Int J Electr Power
Energy Syst 81:254–264
Author Index
A G
Agarwal, Kusum Lata, 251, 335, 343, 357 Gehlot, Kishore Singh, 409
Agarwal, Nitesh, 227 Gupta, Utkarsh, 1, 139
Akter, Morium, 95
Ali, Md. Hayder, 53, 65
Ali, Mohammad Hanif, 53, 65 J
Aseri, Ramesh, 425 Jangid, Aisha, 159
Jha, Aashish Kumar, 111
Jindal, Aditya, 335
Joshi, Ashwani Kumar, 425
B
Joshi, Yogesh, 381
Bhandari, Sanjay, 191
Bhansali, Pratik, 303
Bhati, Tanmay, 449
K
Kanwar, Khamma, 285
Kapoor, Gaurav, 73, 119
C Karia, Megha C., 25
Chaudhary, Laxmi, 159 Khyani, Harish, 295
Choudhary, Arjun, 9 Khyani, Harish Kumar, 199, 219, 449
Choudhary, Ranjana, 397 Kulshrestha, Abhijit, 171
Kumar, Amit, 181
Kumar, Nitesh, 181
D Kushwaha, Ajay, 267
Dave, Sushma, 101
Dhanraj, 251
Dhariwal, Saraswati Chand, 209 L
Lehri, Divam, 9
E M
Eva, Mahmuda Najnin, 95 Mangal, Rajat, 209
Mathur, Akleshwar, 219
Mathur, Akshat, 101
F Mathur, Ashish, 171
Fahmida Islam, Sk., 95 Mathur, Geetika, 171
Ferdous, Jannatul, 95 Mehta, Sandip, 303
https://doi.org/10.1007/978-981-15-8820-4
470 Author Index
Mehta, Vinit, 319, 381 Sharma, Santosh Kumar, 139

Mohanty, A. M., 181 Sharma, Shubham, 343
Sharma, Yatendra Mohan, 133
Shorif, Sumaita Binte, 95
N Shringi, Surbhi, 139
Namjoshi, Sadanand A., 45 Singhal, Amit, 85, 149
Nehiwal, Jayesh, 295 Singh, Bablu Kumar, 191, 277
Singh, Chandershekhar, 267, 295
Singh, Samit Kumar, 45
P Singh, Sudeep Kumar, 181
Panchauli, Dheeraj, 1 Swain, Suvam Sourav, 181
Pandya, Bhavik J., 25
Panwar, Kapil, 335
Patel, Shrawan Ram, 397 T
Patra, Prashanjeet, 181 Tandon, Ankit, 85, 149
Purohit, Rajendra, 227
U
R Uddin, Mohammad Shorif, 95
Ram Patel, Shrawan, 295
Ram, Shrawan, 409
V
Vajpai, Jayashri, 199, 285, 319
S Vishnoi, Shyam Lal, 357
Saini, Pawan Kumar, 133 Vyas, Supriya, 277
Sangani, Kamlesh B., 25
Shalini, 133
Sharma, Komal, 159 Y
Sharma, Nidhi, 111 Yadav, D. K., 1

10.1007@978 981 15 8820 4

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

10.1007@978 981 15 8820 4

Uploaded by

Copyright:

Available Formats

Algorithms for Intelligent Systems

Series Editors: Jagdish Chand Bansal · Kusum Deep · Atulya K. Nagar

More information about this series at http://www.springer.com/series/16171

Kusum Lata Agarwal Mukesh Saraswat

ISSN 2524-7565 ISSN 2524-7573 (electronic)

Twenty-ﬁrst century is the transition from fossil-fuel-based energy to renewable

Dhaka, Bangladesh Mohammad Shorif Uddin

1 Solar PV-Based Electric Vehicle Charging System

9 Vehicle Detection and Its Speed Measurement . . . . . . . . . . . . . . . . 95

22 Renewable Energy as Biofuel from Mirabilis Jalapa Seed Oil . . . . 219

35 Design and Simulation of MPPT-Operated DC-DC Flyback

Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 469

Dr. Mohammad Shorif Uddin is a Professor of Computer Science and

Dr. Avdhesh Sharma is a Professor of Electrical Engineering at M.B.M. Engineering

Dr. Kusum Lata Agarwal is a Professor and Head of Electrical Engineering at

Dr. Mukesh Saraswat is an Associate Professor of Computer Science & Engineering

Utkarsh Gupta, D. K. Yadav, and Dheeraj Panchauli

U. Gupta (B) · D. K. Yadav · D. Panchauli

© Springer Nature Singapore Pte Ltd. 2021 1

A solar PV array of 1 Soltech 1 STH-215-P is used with 10 series modules and 40

2.2 Boost Converter

2.3 Single-Phase Full Bridge Inverter for R-L Load Inverter

A single-phase square wave type voltage source inverter produces square-shaped

Fig.1 Block diagram of the system

Array type: 1Soltech 1STH-215-P; 10 series modules; 40 parallel strings

Fig. 2 The PV and IV characteristics of the PV array are as follows

Fig. 3 The output waveform of the PV array used in this system

Fig. 4 Basic diagram of a boost converter

Fig. 5 Basic diagram of a single-phase inverter

2.4 Single-Phase AC Supply

A resistive load of 5000 ohms is taken in this system.

3 Working of the System

The working of this system is explained basically in three modes.

Fig. 6 Simulation model of the system in Matlab

Fig. 7 Output waveforms across the load

4 Simulation Model and Results

Fig.8 Output waveforms across the battery

Divam Lehri and Arjun Choudhary

NTL = Total Energy Losses(TD) − TL (1)

Total Energy Losses = Energy Supplied − Bills paid (2)

D. Lehri (B) · A. Choudhary

© Springer Nature Singapore Pte Ltd. 2021 9

Table 1 Classification of methods of electricity theft

AT & C Losses = {1 − (BE × CE)} × 100 (3)

T & D Losses = {1 − (BE)} × 100 (4)

Billing efficiency (BE) = Total unit Billed/Total unit Inputs (5)

Collection efficiency (CE) = Revenue collected/Amount Billed (6)

AT & C Losses = TL + CL (7)

cooperation to ensure resiliency, privacy and security of smart grid. A paradigm of

2 Meter Tampering Methodologies

Fig. 2 Actual connection

Fig. 3 Swapping of phase and neutral lines

A summary of mechanisms that are generally used to tamper smart meters is

2.1 Swapping of Phase and Neutral Lines

2.2 Double Feeding

Fig. 4 Double feeding

Fig. 5 Actual condition

2.3 Neutral Missing

2.4 Neutral Disturbance

In this method of tampering some noise (High-Frequency voltage signals) is added

Fig. 6 Neutral missing

Fig. 7 Neutral disturbance

2.5 Current Reversal by Connecting Input and Output